{"cells":[{"cell_type":"markdown","source":"# Uso de spaCY","metadata":{"id":"8UtDTjZ4M3hn","cell_id":"f812aa1400404df6afcbe48d6d562620","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"!python -m spacy download es_core_news_sm\n! pip install -U symspellpy\nimport nltk # importar natural language toolkit\nnltk.download('punkt') \nnltk.download('stopwords') # modulo para descargar stopwords en diferentes idiomas\nnltk.download('wordnet')\nfrom nltk.corpus import stopwords\nimport pandas as pd\nimport numpy as np\nimport re\nimport string\nimport plotly\nimport matplotlib.pyplot as plt\nfrom nltk.stem import PorterStemmer \nimport time\nimport spacy\nimport es_core_news_sm\nfrom nltk.corpus import stopwords\nfrom nltk.tokenize import word_tokenize\nfrom nltk.tokenize import sent_tokenize\nfrom sklearn.feature_extraction.text import CountVectorizer\nfrom sklearn.feature_extraction.text import TfidfTransformer\nfrom nltk.probability import FreqDist\nfrom wordcloud import WordCloud\nimport pickle\nfrom symspellpy import SymSpell\nimport pkg_resources\nfrom symspellpy import SymSpell, Verbosity","metadata":{"id":"qHIBPJTYOBwx","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"4f24c55d9f29402b9473b5a305ae3760","outputId":"e83f5dd3-ba43-4ab1-b468-b9e3e8e7e9ed","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":21095,"user_tz":240,"timestamp":1653425193234},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\nCollecting es_core_news_sm==2.2.5\n Downloading https://github.com/explosion/spacy-models/releases/download/es_core_news_sm-2.2.5/es_core_news_sm-2.2.5.tar.gz (16.2 MB)\n\u001b[K |████████████████████████████████| 16.2 MB 5.2 MB/s \n\u001b[?25hRequirement already satisfied: spacy>=2.2.2 in /usr/local/lib/python3.7/dist-packages (from es_core_news_sm==2.2.5) (2.2.4)\nRequirement already satisfied: cymem<2.1.0,>=2.0.2 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (2.0.6)\nRequirement already satisfied: preshed<3.1.0,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (3.0.6)\nRequirement already satisfied: plac<1.2.0,>=0.9.6 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (1.1.3)\nRequirement already satisfied: wasabi<1.1.0,>=0.4.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (0.9.1)\nRequirement already satisfied: catalogue<1.1.0,>=0.0.7 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (1.0.0)\nRequirement already satisfied: numpy>=1.15.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (1.21.6)\nRequirement already satisfied: requests<3.0.0,>=2.13.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (2.23.0)\nRequirement already satisfied: tqdm<5.0.0,>=4.38.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (4.64.0)\nRequirement already satisfied: murmurhash<1.1.0,>=0.28.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (1.0.7)\nRequirement already satisfied: srsly<1.1.0,>=1.0.2 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (1.0.5)\nRequirement already satisfied: blis<0.5.0,>=0.4.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (0.4.1)\nRequirement already satisfied: thinc==7.4.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (7.4.0)\nRequirement already satisfied: setuptools in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_sm==2.2.5) (57.4.0)\nRequirement already satisfied: importlib-metadata>=0.20 in /usr/local/lib/python3.7/dist-packages (from catalogue<1.1.0,>=0.0.7->spacy>=2.2.2->es_core_news_sm==2.2.5) (4.11.3)\nRequirement already satisfied: zipp>=0.5 in /usr/local/lib/python3.7/dist-packages (from importlib-metadata>=0.20->catalogue<1.1.0,>=0.0.7->spacy>=2.2.2->es_core_news_sm==2.2.5) (3.8.0)\nRequirement already satisfied: typing-extensions>=3.6.4 in /usr/local/lib/python3.7/dist-packages (from importlib-metadata>=0.20->catalogue<1.1.0,>=0.0.7->spacy>=2.2.2->es_core_news_sm==2.2.5) (4.2.0)\nRequirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_sm==2.2.5) (1.24.3)\nRequirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_sm==2.2.5) (3.0.4)\nRequirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_sm==2.2.5) (2.10)\nRequirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_sm==2.2.5) (2022.5.18.1)\nBuilding wheels for collected packages: es-core-news-sm\n Building wheel for es-core-news-sm (setup.py) ... \u001b[?25l\u001b[?25hdone\n Created wheel for es-core-news-sm: filename=es_core_news_sm-2.2.5-py3-none-any.whl size=16172933 sha256=917e7d20c5f7f523b6fd6345df341e0f78a96c11fde073a00548603124eab74b\n Stored in directory: /tmp/pip-ephem-wheel-cache-l6s3j56i/wheels/21/8d/a9/6c1a2809c55dd22cd9644ae503a52ba6206b04aa57ba83a3d8\nSuccessfully built es-core-news-sm\nInstalling collected packages: es-core-news-sm\nSuccessfully installed es-core-news-sm-2.2.5\n\u001b[38;5;2m✔ Download and installation successful\u001b[0m\nYou can now load the model via spacy.load('es_core_news_sm')\nLooking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\nCollecting symspellpy\n Downloading symspellpy-6.7.6-py3-none-any.whl (2.6 MB)\n\u001b[K |████████████████████████████████| 2.6 MB 5.3 MB/s \n\u001b[?25hCollecting editdistpy>=0.1.3\n Downloading editdistpy-0.1.3-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl (125 kB)\n\u001b[K |████████████████████████████████| 125 kB 52.2 MB/s \n\u001b[?25hInstalling collected packages: editdistpy, symspellpy\nSuccessfully installed editdistpy-0.1.3 symspellpy-6.7.6\n[nltk_data] Downloading package punkt to /root/nltk_data...\n[nltk_data] Unzipping tokenizers/punkt.zip.\n[nltk_data] Downloading package stopwords to /root/nltk_data...\n[nltk_data] Unzipping corpora/stopwords.zip.\n[nltk_data] Downloading package wordnet to /root/nltk_data...\n[nltk_data] Unzipping corpora/wordnet.zip.\n"}],"execution_count":6},{"cell_type":"code","source":"!python -m spacy download es_core_news_md","metadata":{"id":"Pd-DOmGVOF0t","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"130b2dcffb22496984b256b90f7f671f","outputId":"a8e321c8-9935-45f0-8186-ffdd23d5a00f","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":12478,"user_tz":240,"timestamp":1653230596954},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Collecting es_core_news_md==2.2.5\n Downloading https://github.com/explosion/spacy-models/releases/download/es_core_news_md-2.2.5/es_core_news_md-2.2.5.tar.gz (78.4 MB)\n\u001b[K |████████████████████████████████| 78.4 MB 1.1 MB/s \n\u001b[?25hRequirement already satisfied: spacy>=2.2.2 in /usr/local/lib/python3.7/dist-packages (from es_core_news_md==2.2.5) (2.2.4)\nRequirement already satisfied: blis<0.5.0,>=0.4.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (0.4.1)\nRequirement already satisfied: tqdm<5.0.0,>=4.38.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (4.64.0)\nRequirement already satisfied: murmurhash<1.1.0,>=0.28.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (1.0.7)\nRequirement already satisfied: setuptools in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (57.4.0)\nRequirement already satisfied: preshed<3.1.0,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (3.0.6)\nRequirement already satisfied: thinc==7.4.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (7.4.0)\nRequirement already satisfied: numpy>=1.15.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (1.21.6)\nRequirement already satisfied: requests<3.0.0,>=2.13.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (2.23.0)\nRequirement already satisfied: wasabi<1.1.0,>=0.4.0 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (0.9.1)\nRequirement already satisfied: srsly<1.1.0,>=1.0.2 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (1.0.5)\nRequirement already satisfied: plac<1.2.0,>=0.9.6 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (1.1.3)\nRequirement already satisfied: cymem<2.1.0,>=2.0.2 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (2.0.6)\nRequirement already satisfied: catalogue<1.1.0,>=0.0.7 in /usr/local/lib/python3.7/dist-packages (from spacy>=2.2.2->es_core_news_md==2.2.5) (1.0.0)\nRequirement already satisfied: importlib-metadata>=0.20 in /usr/local/lib/python3.7/dist-packages (from catalogue<1.1.0,>=0.0.7->spacy>=2.2.2->es_core_news_md==2.2.5) (4.11.3)\nRequirement already satisfied: zipp>=0.5 in /usr/local/lib/python3.7/dist-packages (from importlib-metadata>=0.20->catalogue<1.1.0,>=0.0.7->spacy>=2.2.2->es_core_news_md==2.2.5) (3.8.0)\nRequirement already satisfied: typing-extensions>=3.6.4 in /usr/local/lib/python3.7/dist-packages (from importlib-metadata>=0.20->catalogue<1.1.0,>=0.0.7->spacy>=2.2.2->es_core_news_md==2.2.5) (4.2.0)\nRequirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_md==2.2.5) (2021.10.8)\nRequirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_md==2.2.5) (1.24.3)\nRequirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_md==2.2.5) (3.0.4)\nRequirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.13.0->spacy>=2.2.2->es_core_news_md==2.2.5) (2.10)\n\u001b[38;5;2m✔ Download and installation successful\u001b[0m\nYou can now load the model via spacy.load('es_core_news_md')\n"}],"execution_count":null},{"cell_type":"code","source":"import es_core_news_md\nnlp = es_core_news_md.load()","metadata":{"id":"vsxa_DTEMtNC","cell_id":"23d2d20f6f154d72aa5e1533fdcc3ce8","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"# Leer un string","metadata":{"id":"2uB1-2VbOeRt","cell_id":"7479163554e64be5bbb99dd13c66b8ed","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"texto= ('Este es un tutorial acerca de Procesamiento de lenguaje usando Python con spaCy')\ndoc = nlp(texto)\n#tokenizar\nprint([token.text for token in doc])","metadata":{"id":"ht5U8hbEM6pA","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"d690cc6ee3c74a9fbb5fd3742f1ec177","outputId":"0e31fd03-3c5f-4fa0-8f08-20ce28ca5783","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":587,"user_tz":240,"timestamp":1653230907520},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"['Este', 'es', 'un', 'tutorial', 'acerca', 'de', 'Procesamiento', 'de', 'lenguaje', 'usando', 'Python', 'con', 'spaCy']\n"}],"execution_count":null},{"cell_type":"markdown","source":"# Deteccion de oraciones","metadata":{"id":"eB-FMmvvOh90","cell_id":"b7abf56c76714d4ba61c87523d0862d8","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"text=('Gus es un desarrollador en Python actualmente trabajando para una compañia Fintech en Londres Inglaterra. Se encuentra interesado en aprender NLP.')\nt=nlp(text)\noraciones= list(t.sents)\nprint(len(oraciones))\nfor x in oraciones:\n print(x)","metadata":{"id":"teHetL6wOmGc","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"5a87d51bfdd6417687ec454fdecbd186","outputId":"32f4d97d-6355-4adc-8433-721706c7b3e0","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":416,"user_tz":240,"timestamp":1653230960361},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"2\nGus es un desarrollador en Python actualmente trabajando para una compañia Fintech en Londres Inglaterra.\nSe encuentra interesado en aprender NLP.\n"}],"execution_count":null},{"cell_type":"markdown","source":"# Tokenizacion","metadata":{"id":"RrNKtx8EPrr1","cell_id":"727263397ef84421a4722da4ad36ea8c","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"for token in t:\n print(token, token.idx)","metadata":{"id":"RROu30nCPs11","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"e1ad0b4edc734566b9a6b20f88fce279","outputId":"4773ec60-bee9-459f-86be-b230e7965ea5","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":415,"user_tz":240,"timestamp":1653231007774},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Gus 0\nes 4\nun 7\ndesarrollador 10\nen 24\nPython 27\nactualmente 34\ntrabajando 46\npara 57\nuna 62\ncompañia 66\nFintech 75\nen 83\nLondres 86\nInglaterra 94\n. 104\nSe 106\nencuentra 109\ninteresado 119\nen 130\naprender 133\nNLP 142\n. 145\n"}],"execution_count":null},{"cell_type":"code","source":"for token in t:\n print(token, token.idx, token.text_with_ws,\n token.is_alpha, token.is_punct, token.is_space,\n token.shape_, token.is_stop)","metadata":{"id":"Rhv5on_BP9Jt","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"a9c420ace2cc4b3f90253afa50ab9128","outputId":"5e73a8b8-c0a6-44b5-89dc-5432c9db9071","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":404,"user_tz":240,"timestamp":1653231066952},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Gus 0 Gus True False False Xxx False\nes 4 es True False False xx True\nun 7 un True False False xx True\ndesarrollador 10 desarrollador True False False xxxx False\nen 24 en True False False xx True\nPython 27 Python True False False Xxxxx False\nactualmente 34 actualmente True False False xxxx True\ntrabajando 46 trabajando True False False xxxx False\npara 57 para True False False xxxx True\nuna 62 una True False False xxx True\ncompañia 66 compañia True False False xxxx False\nFintech 75 Fintech True False False Xxxxx False\nen 83 en True False False xx True\nLondres 86 Londres True False False Xxxxx False\nInglaterra 94 Inglaterra True False False Xxxxx False\n. 104 . False True False . False\nSe 106 Se True False False Xx True\nencuentra 109 encuentra True False False xxxx True\ninteresado 119 interesado True False False xxxx False\nen 130 en True False False xx True\naprender 133 aprender True False False xxxx False\nNLP 142 NLP True False False XXX False\n. 145 . False True False . False\n"}],"execution_count":null},{"cell_type":"markdown","source":"En este ejemplo tenemos:\n\n- text_with_ws imprime el texto del token \n- is_alpha detecta si el token consiste en caracteres alfa numericos o no\n- is_punct detecta si el token es un simbolo de puntuacion.\n- is_space detecta si el token es un espacio o no.\n- shape_ imprime el output shape de la palabra.\n- is_stop detecta si es una stopword o no","metadata":{"id":"v6FvUx-UQFoz","cell_id":"042720b31c574d13a3539ffcefc017b0","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Stopwords","metadata":{"id":"Lq1Cev75Qjy5","cell_id":"f45428c7ff7e4b66b7c110958ae403cd","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"import spacy\nspacy_stopwords = spacy.lang.es.stop_words.STOP_WORDS\nprint(len(spacy_stopwords))\nfor stop_word in list(spacy_stopwords)[:10]:\n print(stop_word)","metadata":{"id":"_h2N2XmmQmaC","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"bd1bcf583a5446a3835b8b8a454fb372","outputId":"329adf96-1b13-4b92-8c1e-93b3d6df645d","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":432,"user_tz":240,"timestamp":1653231241658},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"551\nsea\nsiguiente\ninformo\ntampoco\ndan\nestuvo\nenseguida\nhabia\nninguno\nocho\n"}],"execution_count":null},{"cell_type":"code","source":"for token in t:\n if not token.is_stop:\n print(token)","metadata":{"id":"PF9yA7o7Qu-5","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"a32eb3981fb2483a82695ae4ffaf803e","outputId":"296aa79f-e135-4fd6-e33f-ea5e2b5cea2d","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":516,"user_tz":240,"timestamp":1653231266950},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Gus\ndesarrollador\nPython\ntrabajando\ncompañia\nFintech\nLondres\nInglaterra\n.\ninteresado\naprender\nNLP\n.\n"}],"execution_count":null},{"cell_type":"code","source":"# Creacion adicional de stopwrods\ndocumento_sin_stopword = [token for token in t if not token.is_stop]\nprint(documento_sin_stopword)","metadata":{"id":"XZqDOi_qQ1td","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"a3ef3deb5dea4f2e95b6d260184220cc","outputId":"2e6b41a6-6e11-43a9-ab8b-2c7b670bdf3e","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":403,"user_tz":240,"timestamp":1653231317027},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"[Gus, desarrollador, Python, trabajando, compañia, Fintech, Londres, Inglaterra, ., interesado, aprender, NLP, .]\n"}],"execution_count":null},{"cell_type":"markdown","source":"# Lemantizacion","metadata":{"id":"27Owz6clRAe2","cell_id":"253eae6ddd004a6c8a174b286075193f","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"for token in t:\n print(token, '-', token.lemma_)","metadata":{"id":"siLYeWXjRDl6","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"71a15384d0e742dcbbb651be6f28ad8e","outputId":"c6af8fa3-0409-402b-95a1-c5e256ea694b","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":619,"user_tz":240,"timestamp":1653231365055},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Gus - Gus\nes - ser\nun - uno\ndesarrollador - desarrollador\nen - en\nPython - Python\nactualmente - actualmente\ntrabajando - trabajar\npara - parir\nuna - uno\ncompañia - compañia\nFintech - Fintech\nen - en\nLondres - Londres\nInglaterra - Inglaterra\n. - .\nSe - Se\nencuentra - encontrar\ninteresado - interesar\nen - en\naprender - aprender\nNLP - NLP\n. - .\n"}],"execution_count":null},{"cell_type":"markdown","source":"# Word Frequency","metadata":{"id":"WPyUzf66RL-V","cell_id":"aa286db56dbc4a09a089973a71be8feb","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"texto= '''\nLa FIFA responde así a una denuncia interpuesta por la Federación de Chile ante esa Comisión Disciplinaria, en la que presentaba alegaciones sobre la posible falsificación de los documentos que conceden la nacionalidad ecuatoriana Byron Castillo.\n\nLa selección de Ecuador se clasificó de forma directa para el Mundial, junto con las de Brasil, Argentina y Uruguay, al contrario que las de Chile y Perú. El combinado peruano, que terminó quinto por detrás del ecuatoriano, disputará una repesca.\n\nEl defensa fue alineado por el seleccionador ecuatoriano Gustavo Alfaro para los dos partidos contra Paraguay y Chile y en una ocasión ante Uruguay, Bolivia, Venezuela y Argentina, partidos clave para que el equipo lograse uno de los cupos directos para el Mundial.\n\n\"Innumerables pruebas de que nació en Colombia\"\nLa Federación de Chile denunció el pasado día 5 que hay \"innumerables pruebas de que el jugador nació en Colombia\".\n\n\"Las investigaciones realizadas en Ecuador, entre ellas, un informe jurídico de la Dirección Nacional de Registro Civil, declararon la existencia de inconsistencias en el certificado de nacimiento presentado por el jugador\", afirmó este organismo, que acusó a la Federación Ecuatoriana de tener \"total conocimiento\" de las irregularidades.\n\nUna posible sanción de la FIFA podría implicar la resta de puntos a Ecuador por los partidos que Castillo jugó, lo que alteraría la nómina de clasificados.\n\nUn informe técnico jurídico de la dirección nacional del registro civil de Ecuador afirma que la inscripción de nacimiento de Byron Castillo en la ciudad ecuatoriana de Guayas no consta en el tomo, la página y el acta solicitado, según un documento oficial.\n'''","metadata":{"id":"a_P6U9n1RNUo","cell_id":"f324339c05264984a0ef0c7e07d866e4","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"type(texto)","metadata":{"id":"OVFSTmTCSbUI","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"02432d1e992341a19bb81acbfc9cd47c","outputId":"d57fe1df-0380-4867-96cd-c117691fe361","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":3,"user_tz":240,"timestamp":1653231750249},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"str"},"metadata":{},"execution_count":49}],"execution_count":null},{"cell_type":"code","source":"import re\ntexto1 = re.sub('\\n', '', texto) #remover saltos de linea\nprint(type(texto))\nstr(texto1)","metadata":{"id":"YJEqi4TuSNHZ","colab":{"height":123,"base_uri":"https://localhost:8080/"},"cell_id":"a72b0b608d7a4393930f95dc1ac1ea55","outputId":"a2e6d4b3-f4e7-4dbc-b305-aeb48c854f25","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":389,"user_tz":240,"timestamp":1653231764290},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"\n"},{"output_type":"execute_result","data":{"text/plain":"'La FIFA responde así a una denuncia interpuesta por la Federación de Chile ante esa Comisión Disciplinaria, en la que presentaba alegaciones sobre la posible falsificación de los documentos que conceden la nacionalidad ecuatoriana Byron Castillo.La selección de Ecuador se clasificó de forma directa para el Mundial, junto con las de Brasil, Argentina y Uruguay, al contrario que las de Chile y Perú. El combinado peruano, que terminó quinto por detrás del ecuatoriano, disputará una repesca.El defensa fue alineado por el seleccionador ecuatoriano Gustavo Alfaro para los dos partidos contra Paraguay y Chile y en una ocasión ante Uruguay, Bolivia, Venezuela y Argentina, partidos clave para que el equipo lograse uno de los cupos directos para el Mundial.\"Innumerables pruebas de que nació en Colombia\"La Federación de Chile denunció el pasado día 5 que hay \"innumerables pruebas de que el jugador nació en Colombia\".\"Las investigaciones realizadas en Ecuador, entre ellas, un informe jurídico de la Dirección Nacional de Registro Civil, declararon la existencia de inconsistencias en el certificado de nacimiento presentado por el jugador\", afirmó este organismo, que acusó a la Federación Ecuatoriana de tener \"total conocimiento\" de las irregularidades.Una posible sanción de la FIFA podría implicar la resta de puntos a Ecuador por los partidos que Castillo jugó, lo que alteraría la nómina de clasificados.Un informe técnico jurídico de la dirección nacional del registro civil de Ecuador afirma que la inscripción de nacimiento de Byron Castillo en la ciudad ecuatoriana de Guayas no consta en el tomo, la página y el acta solicitado, según un documento oficial.'","application/vnd.google.colaboratory.intrinsic+json":{"type":"string"}},"metadata":{},"execution_count":51}],"execution_count":null},{"cell_type":"code","source":"doc= nlp(texto1)\n# Remover stopwrods\nwords= [token.text for token in doc if not token.is_stop and not token.is_punct]\nfrom collections import Counter\nword_freq= Counter(words)\n# Sacar las 5 mas frecuentes y sus frecuencias\ncommon_words= word_freq.most_common(5)\nprint(common_words)\nunique_words = [word for (word, freq) in word_freq.items() if freq == 1]\nprint(unique_words)","metadata":{"id":"BaFYahOoRd6h","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"920c69b3acfe4770a74135abcfca471e","outputId":"7b82df60-ad71-4ef2-f8ad-a8d2856c8bdf","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":409,"user_tz":240,"timestamp":1653231850729},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"[('y', 6), ('Chile', 4), ('Ecuador', 4), ('a', 3), ('Federación', 3)]\n['responde', 'denuncia', 'interpuesta', 'Comisión', 'Disciplinaria', 'presentaba', 'alegaciones', 'falsificación', 'documentos', 'conceden', 'nacionalidad', 'selección', 'clasificó', 'forma', 'directa', 'Brasil', 'contrario', 'Perú', 'combinado', 'peruano', 'terminó', 'quinto', 'disputará', 'repesca', 'defensa', 'alineado', 'seleccionador', 'Gustavo', 'Alfaro', 'Paraguay', 'ocasión', 'Bolivia', 'Venezuela', 'clave', 'equipo', 'lograse', 'cupos', 'directos', '\"Innumerables', 'Colombia\"La', 'denunció', '5', 'innumerables', 'Colombia\"', '\"Las', 'investigaciones', 'realizadas', 'Dirección', 'Nacional', 'Registro', 'Civil', 'declararon', 'existencia', 'inconsistencias', 'certificado', 'presentado', 'organismo', 'acusó', 'Ecuatoriana', 'conocimiento', 'irregularidades', 'sanción', 'implicar', 'resta', 'puntos', 'jugó', 'alteraría', 'nómina', 'clasificados', 'técnico', 'dirección', 'nacional', 'registro', 'civil', 'afirma', 'inscripción', 'ciudad', 'Guayas', 'consta', 'tomo', 'página', 'acta', 'solicitado', 'documento', 'oficial']\n"}],"execution_count":null},{"cell_type":"markdown","source":"# POS Tagging","metadata":{"id":"qzhKZ2n_TEv4","cell_id":"8e43790071cc404cae6d2d56eebad3b2","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"for token in doc:\n print(token,' -', token.tag_, ' -', token.pos_,' -' ,spacy.explain(token.tag_))","metadata":{"id":"KEF70sIAS-1d","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"5c90b241073840da952f4eb39bddd272","outputId":"a023495b-4956-47f3-e997-77899a3ecedd","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":383,"user_tz":240,"timestamp":1653231956391},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"La - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nFIFA - PROPN___ - PROPN - None\nresponde - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin - VERB - None\nasí - ADV___ - ADV - None\na - ADP__AdpType=Prep - ADP - None\nuna - DET__Definite=Ind|Gender=Fem|Number=Sing|PronType=Art - DET - None\ndenuncia - NOUN__Gender=Fem|Number=Sing - NOUN - None\ninterpuesta - ADJ__Gender=Fem|Number=Sing|VerbForm=Part - ADJ - None\npor - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nFederación - PROPN___ - PROPN - None\nde - ADP__AdpType=Prep - ADP - None\nChile - PROPN___ - PROPN - None\nante - ADP__AdpType=Prep - ADP - None\nesa - DET__Gender=Fem|Number=Sing|PronType=Dem - DET - None\nComisión - PROPN___ - PROPN - None\nDisciplinaria - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nen - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nque - PRON__PronType=Rel - PRON - None\npresentaba - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Imp|VerbForm=Fin - VERB - None\nalegaciones - NOUN__Gender=Fem|Number=Plur - NOUN - None\nsobre - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nposible - ADJ__Number=Sing - ADJ - None\nfalsificación - NOUN__Gender=Fem|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nlos - DET__Definite=Def|Gender=Masc|Number=Plur|PronType=Art - DET - None\ndocumentos - NOUN__Gender=Masc|Number=Plur - NOUN - None\nque - PRON__PronType=Rel - PRON - None\nconceden - VERB__Mood=Ind|Number=Plur|Person=3|Tense=Pres|VerbForm=Fin - VERB - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nnacionalidad - NOUN__Gender=Fem|Number=Sing - NOUN - None\necuatoriana - ADJ__Gender=Fem|Number=Sing - ADJ - None\nByron - PROPN___ - PROPN - None\nCastillo - PROPN___ - PROPN - None\n. - PUNCT__PunctType=Peri - PUNCT - None\nLa - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nselección - NOUN__Gender=Fem|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nEcuador - PROPN___ - PROPN - None\nse - PRON__Person=3 - PRON - None\nclasificó - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\nde - ADP__AdpType=Prep - ADP - None\nforma - NOUN__Gender=Fem|Number=Sing - NOUN - None\ndirecta - ADJ__Gender=Fem|Number=Sing - ADJ - None\npara - ADP__AdpType=Prep - ADP - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\nMundial - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\njunto - ADJ__AdpType=Prep - ADJ - None\ncon - ADP__AdpType=Prep - ADP - None\nlas - DET__Definite=Def|Gender=Fem|Number=Plur|PronType=Art - DET - None\nde - ADP__AdpType=Prep - ADP - None\nBrasil - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nArgentina - PROPN___ - PROPN - None\ny - CCONJ___ - CONJ - None\nUruguay - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nal - ADP__AdpType=Preppron|Gender=Masc|Number=Sing - ADP - None\ncontrario - NOUN___ - NOUN - None\nque - SCONJ___ - SCONJ - None\nlas - DET__Definite=Def|Gender=Fem|Number=Plur|PronType=Art - DET - None\nde - ADP__AdpType=Prep - ADP - None\nChile - PROPN___ - PROPN - None\ny - CCONJ___ - CONJ - None\nPerú - PROPN___ - PROPN - None\n. - PUNCT__PunctType=Peri - PUNCT - None\nEl - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\ncombinado - NOUN__Gender=Masc|Number=Sing - NOUN - None\nperuano - ADJ__Gender=Masc|Number=Sing - ADJ - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nque - PRON__PronType=Rel - PRON - None\nterminó - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\nquinto - ADJ__Gender=Masc|Number=Sing|NumType=Ord - ADJ - None\npor - ADP__AdpType=Prep - ADP - None\ndetrás - ADV__AdpType=Preppron|Gender=Masc|Number=Sing - ADV - None\ndel - ADP__AdpType=Preppron|Gender=Masc|Number=Sing - ADP - None\necuatoriano - NOUN__Gender=Masc|Number=Sing - NOUN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\ndisputará - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Fut|VerbForm=Fin - VERB - None\nuna - DET__Definite=Ind|Gender=Fem|Number=Sing|PronType=Art - DET - None\nrepesca - NOUN__Gender=Fem|Number=Sing - NOUN - None\n. - PUNCT__PunctType=Peri - PUNCT - None\nEl - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\ndefensa - NOUN__Number=Sing - NOUN - None\nfue - AUX__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - AUX - None\nalineado - VERB__Gender=Masc|Number=Sing|Tense=Past|VerbForm=Part - VERB - None\npor - ADP__AdpType=Prep - ADP - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\nseleccionador - NOUN__Gender=Masc|Number=Sing - NOUN - None\necuatoriano - ADJ__Gender=Masc|Number=Sing - ADJ - None\nGustavo - PROPN___ - PROPN - None\nAlfaro - PROPN___ - PROPN - None\npara - ADP__AdpType=Prep - ADP - None\nlos - DET__Definite=Def|Gender=Masc|Number=Plur|PronType=Art - DET - None\ndos - NUM__Number=Plur|NumType=Card - NUM - None\npartidos - NOUN__Gender=Masc|Number=Plur - NOUN - None\ncontra - ADP__AdpType=Prep - ADP - None\nParaguay - PROPN___ - PROPN - None\ny - CCONJ___ - CONJ - None\nChile - PROPN___ - PROPN - None\ny - CCONJ___ - CONJ - None\nen - ADP__AdpType=Prep - ADP - None\nuna - DET__Definite=Ind|Gender=Fem|Number=Sing|PronType=Art - DET - None\nocasión - NOUN__Gender=Fem|Number=Sing - NOUN - None\nante - ADP__AdpType=Prep - ADP - None\nUruguay - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nBolivia - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nVenezuela - PROPN___ - PROPN - None\ny - CCONJ___ - CONJ - None\nArgentina - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\npartidos - NOUN__Gender=Masc|Number=Plur - NOUN - None\nclave - NOUN__Gender=Fem|Number=Sing - NOUN - None\npara - ADP__AdpType=Prep - ADP - None\nque - SCONJ___ - SCONJ - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\nequipo - NOUN__Gender=Masc|Number=Sing - NOUN - None\nlograse - VERB__Mood=Sub|Number=Sing|Person=3|Tense=Imp|VerbForm=Fin - VERB - None\nuno - PRON__Gender=Masc|Number=Sing|PronType=Ind - PRON - None\nde - ADP__AdpType=Prep - ADP - None\nlos - DET__Definite=Def|Gender=Masc|Number=Plur|PronType=Art - DET - None\ncupos - NOUN__Gender=Masc|Number=Plur - NOUN - None\ndirectos - ADJ__Gender=Masc|Number=Plur - ADJ - None\npara - ADP__AdpType=Prep - ADP - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\nMundial - PROPN___ - PROPN - None\n. - PUNCT__PunctType=Peri - PUNCT - None\n\"Innumerables - ADJ__Number=Plur - ADJ - None\npruebas - NOUN__Gender=Fem|Number=Plur - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nque - SCONJ___ - SCONJ - None\nnació - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\nen - ADP__AdpType=Prep - ADP - None\nColombia\"La - PROPN___ - PROPN - None\nFederación - PROPN___ - PROPN - None\nde - ADP__AdpType=Prep - ADP - None\nChile - PROPN___ - PROPN - None\ndenunció - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\npasado - ADJ__Gender=Masc|Number=Sing|VerbForm=Part - ADJ - None\ndía - NOUN___ - NOUN - None\n5 - NUM__NumForm=Digit|NumType=Card - NUM - None\nque - PRON__PronType=Rel - PRON - None\nhay - AUX__Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin - AUX - None\n\" - PUNCT__PunctType=Quot - PUNCT - None\ninnumerables - ADJ__Number=Plur - ADJ - None\npruebas - NOUN__Gender=Fem|Number=Plur - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nque - SCONJ___ - SCONJ - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\njugador - NOUN__Gender=Masc|Number=Sing - NOUN - None\nnació - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\nen - ADP__AdpType=Prep - ADP - None\nColombia\" - PROPN___ - PROPN - None\n. - PUNCT__PunctType=Peri - PUNCT - None\n\"Las - NUM__NumForm=Digit - NUM - None\ninvestigaciones - NOUN__Gender=Fem|Number=Plur - NOUN - None\nrealizadas - ADJ__Gender=Fem|Number=Plur|VerbForm=Part - ADJ - None\nen - ADP__AdpType=Prep - ADP - None\nEcuador - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nentre - ADP__AdpType=Prep - ADP - None\nellas - PRON__Gender=Fem|Number=Plur|Person=3|PronType=Prs - PRON - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nun - DET__Definite=Ind|Gender=Masc|Number=Sing|PronType=Art - DET - None\ninforme - NOUN__Gender=Masc|Number=Sing - NOUN - None\njurídico - ADJ__Gender=Masc|Number=Sing - ADJ - None\nde - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nDirección - PROPN___ - PROPN - None\nNacional - PROPN___ - PROPN - None\nde - ADP__AdpType=Prep - ADP - None\nRegistro - PROPN___ - PROPN - None\nCivil - PROPN___ - PROPN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\ndeclararon - VERB__Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin - VERB - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nexistencia - NOUN__Gender=Fem|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\ninconsistencias - NOUN__Gender=Fem|Number=Plur - NOUN - None\nen - ADP__AdpType=Prep - ADP - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\ncertificado - NOUN__Gender=Masc|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nnacimiento - NOUN__Gender=Masc|Number=Sing - NOUN - None\npresentado - ADJ__Gender=Masc|Number=Sing|VerbForm=Part - ADJ - None\npor - ADP__AdpType=Prep - ADP - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\njugador - NOUN__Gender=Masc|Number=Sing - NOUN - None\n\" - PUNCT__PunctType=Quot - PUNCT - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nafirmó - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\neste - DET__Gender=Masc|Number=Sing|PronType=Dem - DET - None\norganismo - NOUN__Gender=Masc|Number=Sing - NOUN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nque - PRON__PronType=Rel - PRON - None\nacusó - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\na - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nFederación - PROPN___ - PROPN - None\nEcuatoriana - PROPN___ - PROPN - None\nde - ADP__AdpType=Prep - ADP - None\ntener - VERB__VerbForm=Inf - VERB - None\n\" - PUNCT__PunctType=Quot - PUNCT - None\ntotal - ADJ__Number=Sing - ADJ - None\nconocimiento - NOUN__Gender=Masc|Number=Sing - NOUN - None\n\" - PUNCT__PunctType=Quot - PUNCT - None\nde - ADP__AdpType=Prep - ADP - None\nlas - DET__Definite=Def|Gender=Fem|Number=Plur|PronType=Art - DET - None\nirregularidades - NOUN__Gender=Fem|Number=Plur - NOUN - None\n. - PUNCT__PunctType=Peri - PUNCT - None\nUna - DET__Definite=Ind|Gender=Fem|Number=Sing|PronType=Art - DET - None\nposible - ADJ__Number=Sing - ADJ - None\nsanción - NOUN__Gender=Fem|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nFIFA - PROPN___ - PROPN - None\npodría - AUX__Mood=Cnd|Number=Sing|Person=3|VerbForm=Fin - AUX - None\nimplicar - VERB__VerbForm=Inf - VERB - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nresta - NOUN__Gender=Fem|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\npuntos - NOUN__Gender=Masc|Number=Plur - NOUN - None\na - ADP__AdpType=Prep - ADP - None\nEcuador - PROPN___ - PROPN - None\npor - ADP__AdpType=Prep - ADP - None\nlos - DET__Definite=Def|Gender=Masc|Number=Plur|PronType=Art - DET - None\npartidos - NOUN__Gender=Masc|Number=Plur - NOUN - None\nque - PRON__PronType=Rel - PRON - None\nCastillo - PROPN___ - PROPN - None\njugó - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin - VERB - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nlo - DET__Definite=Def|Number=Sing|PronType=Art - DET - None\nque - PRON__PronType=Rel - PRON - None\nalteraría - VERB__Mood=Cnd|Number=Sing|Person=3|VerbForm=Fin - VERB - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nnómina - NOUN__Gender=Fem|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nclasificados - NOUN__Gender=Masc|Number=Plur - NOUN - None\n. - PUNCT__PunctType=Peri - PUNCT - None\nUn - DET__Definite=Ind|Gender=Masc|Number=Sing|PronType=Art - DET - None\ninforme - NOUN__Gender=Masc|Number=Sing - NOUN - None\ntécnico - ADJ__Gender=Masc|Number=Sing - ADJ - None\njurídico - ADJ__Gender=Masc|Number=Sing - ADJ - None\nde - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\ndirección - PROPN___ - PROPN - None\nnacional - PROPN___ - PROPN - None\ndel - ADP__AdpType=Preppron|Gender=Masc|Number=Sing - ADP - None\nregistro - PROPN___ - PROPN - None\ncivil - ADJ__Number=Sing - ADJ - None\nde - ADP__AdpType=Prep - ADP - None\nEcuador - PROPN___ - PROPN - None\nafirma - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin - VERB - None\nque - SCONJ___ - SCONJ - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\ninscripción - NOUN__Gender=Fem|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nnacimiento - NOUN__Gender=Masc|Number=Sing - NOUN - None\nde - ADP__AdpType=Prep - ADP - None\nByron - PROPN___ - PROPN - None\nCastillo - PROPN___ - PROPN - None\nen - ADP__AdpType=Prep - ADP - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\nciudad - NOUN__Gender=Fem|Number=Sing - NOUN - None\necuatoriana - ADJ__Gender=Fem|Number=Sing - ADJ - None\nde - ADP__AdpType=Prep - ADP - None\nGuayas - PROPN___ - PROPN - None\nno - ADV__Polarity=Neg - ADV - None\nconsta - VERB__Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin - VERB - None\nen - ADP__AdpType=Prep - ADP - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\ntomo - NOUN__Gender=Masc|Number=Sing - NOUN - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nla - DET__Definite=Def|Gender=Fem|Number=Sing|PronType=Art - DET - None\npágina - NOUN__Gender=Fem|Number=Sing - NOUN - None\ny - CCONJ___ - CONJ - None\nel - DET__Definite=Def|Gender=Masc|Number=Sing|PronType=Art - DET - None\nacta - NOUN__Gender=Masc|Number=Sing - NOUN - None\nsolicitado - ADJ__Gender=Masc|Number=Sing|VerbForm=Part - ADJ - None\n, - PUNCT__PunctType=Comm - PUNCT - None\nsegún - ADP__AdpType=Prep - ADP - None\nun - DET__Definite=Ind|Gender=Masc|Number=Sing|PronType=Art - DET - None\ndocumento - NOUN__Gender=Masc|Number=Sing - NOUN - None\noficial - ADJ__Number=Sing - ADJ - None\n. - PUNCT__PunctType=Peri - PUNCT - None\n"}],"execution_count":null},{"cell_type":"markdown","source":"1. tag_ listas las estructuras finas del POS\n2. pos_ lista la estructuras gruesas del POS\n3. spacy.explain da los detalles descriptivos sobre un POS particular","metadata":{"id":"A9U1VToeTcGl","cell_id":"23c45601576a44b984be3beb2edd468f","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"nouns=[]\nadjectives=[]\nfor token in doc:\n if token.pos_ =='NOUN':\n nouns.append(token)\n if token.pos_ =='ADJ':\n adjectives.append(token)\nprint(nouns)\nprint(adjectives)","metadata":{"id":"ePn4C2N_ToHG","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"d7ea3dc883974e9a9bc94b698254481b","outputId":"df093f04-ee60-4f07-f1c6-c6d61653ef4b","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":387,"user_tz":240,"timestamp":1653232061050},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"[denuncia, alegaciones, falsificación, documentos, nacionalidad, selección, forma, contrario, combinado, ecuatoriano, repesca, defensa, seleccionador, partidos, ocasión, partidos, clave, equipo, cupos, pruebas, día, pruebas, jugador, investigaciones, informe, existencia, inconsistencias, certificado, nacimiento, jugador, organismo, conocimiento, irregularidades, sanción, resta, puntos, partidos, nómina, clasificados, informe, inscripción, nacimiento, ciudad, tomo, página, acta, documento]\n[interpuesta, posible, ecuatoriana, directa, junto, peruano, quinto, ecuatoriano, directos, \"Innumerables, pasado, innumerables, realizadas, jurídico, presentado, total, posible, técnico, jurídico, civil, ecuatoriana, solicitado, oficial]\n"}],"execution_count":null},{"cell_type":"markdown","source":"# Visualizacion usando spaCY","metadata":{"id":"asrthaXMT1j0","cell_id":"b25482a05f2042c2ae723fef9488cbd4","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"from spacy import displacy\ntexto = ('el se encuentra interesado en aprender Procesamiento de Lenguaje Natural')\nt = nlp(texto)\ndisplacy.render(t, style='dep',jupyter=True)","metadata":{"id":"67YSTloeT3Ou","colab":{"height":354,"base_uri":"https://localhost:8080/"},"cell_id":"770eaf8a2d7f4913b94b2cc674f677a1","outputId":"9d9d409c-4475-45d8-bbc1-654576b1678a","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":424,"user_tz":240,"timestamp":1653232166048},"deepnote_cell_type":"code"},"outputs":[{"output_type":"display_data","data":{"text/plain":"","text/html":"\n\n el\n DET\n\n\n\n se\n PRON\n\n\n\n encuentra\n AUX\n\n\n\n interesado\n ADJ\n\n\n\n en\n ADP\n\n\n\n aprender\n VERB\n\n\n\n Procesamiento\n PROPN\n\n\n\n de\n ADP\n\n\n\n Lenguaje\n NOUN\n\n\n\n Natural\n PROPN\n\n\n\n \n \n det\n \n \n\n\n\n \n \n obj\n \n \n\n\n\n \n \n cop\n \n \n\n\n\n \n \n mark\n \n \n\n\n\n \n \n xcomp\n \n \n\n\n\n \n \n obj\n \n \n\n\n\n \n \n case\n \n \n\n\n\n \n \n nmod\n \n \n\n\n\n \n \n flat\n \n \n\n"},"metadata":{}}],"execution_count":null},{"cell_type":"markdown","source":"# Caso aplicado: ¿Cómo podemos predecir el sentimiento asociado con una interacción con el cliente?","metadata":{"id":"Tkpc-IyM6Vv7","cell_id":"36349b8c574c4eac948fb4d016d281aa","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"import os\nfrom collections import Counter\n\nimport matplotlib.pyplot as plt\nimport nltk\nimport numpy as np\nimport pandas as pd\n\nfrom pylab import rcParams\nfrom wordcloud import WordCloud\nfrom nltk import word_tokenize\n\nfrom sklearn.feature_extraction.text import CountVectorizer\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.metrics import accuracy_score\nfrom sklearn.metrics import f1_score\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.feature_extraction.text import TfidfVectorizer\n\nrcParams['figure.figsize'] = 30, 60\n%matplotlib inline","metadata":{"id":"eY8dg6Hc6e4m","cell_id":"1e59ea0ff566444dbd7ac0b911f0ce83","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"# Metas\n\nDespués de trabajar en este caso, debería poder aplicar el análisis de sentimientos a los problemas comerciales. Comprenderá cómo utilizar técnicas de clasificación de texto para crear un modelo de análisis de opiniones y podrá aplicar modelos de opiniones a datos del mundo real.\n\nHabrá adquirido experiencia en el uso de varias técnicas diferentes de vectorización y creación de modelos, utilizando bibliotecas populares de Python como scikit-learn y gensim. Habrá utilizado tanto vectorizaciones basadas en conteo como incrustaciones de palabras, y comprenderá los pros y los contras de cada una.","metadata":{"id":"3wPhXiFl6kbd","cell_id":"930b83720cfe475ba575c4e9b717d214","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Introducción\n**Contexto empresarial**. Eres un científico de datos para una gran empresa de comercio electrónico. Tienes decenas de miles de clientes que escriben reseñas sobre productos cada día. Cada revisión contiene comentarios textuales junto con un sistema de calificación de 1 a 5 estrellas (siendo 1 la menos satisfecha y 5 la más satisfecha). \n\nTambién tiene un equipo de atención al cliente que interactúa con los clientes a través de servicios de llamadas y mensajes. Su empresa también recopila comentarios sobre las experiencias de sus clientes con la interacción del sitio web después de cada compra. Ni este comentario ni el servicio de mensajería tienen un número de calificación. La empresa quiere cuantificar la satisfacción del cliente proveniente de estas interacciones no calificadas para ayudar con futuras decisiones comerciales (por ejemplo, determinar qué tan bien se están desempeñando sus diversos agentes de servicio al cliente).\n\n**Problema de negocio** Su tarea es construir modelos que puedan identificar el sentimiento (positivo o negativo) de cada una de estas interacciones no calificadas.\n\n**Contexto analitico** Los datos son un conjunto de reseñas en formato de archivo CSV. Combinaremos lo que aprendimos sobre el procesamiento de texto y los modelos de clasificación para desarrollar algoritmos capaces de clasificar las interacciones por sentimiento.\n\nEl caso está estructurado de la siguiente manera: 1) leerá y analizará los datos del texto de entrada y las variables de respuesta correspondientes (calificaciones); 2) realizara un preprocesamiento básico para preparar los datos para el modelado; 3) aprendera y aplicara varias formas de caracterizar el texto de reseñas; y finalmente 4) construira modelos de aprendizaje automático para clasificar el texto como mostrando un sentimiento positivo o negativo (1 o 0).","metadata":{"id":"KMDBHuy06rc1","cell_id":"fc54e374072049c393111162674d8f0d","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Lectura de datos y analisis basico","metadata":{"id":"hhEWNyxp7J9a","cell_id":"772a2b5a3af24249a12970af737c8e7b","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"from google.colab import drive\nimport os\ndrive.mount('/content/gdrive')\n# Establecer ruta de acceso en dr\nimport os\nprint(os.getcwd())\nos.chdir(\"/content/gdrive/My Drive\")","metadata":{"id":"CQFxWy6n7Wda","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"2680cc27e94e4058a2dcc1170f0d9379","outputId":"02a015f7-e18f-4eaf-d6c3-e3eaf49c6974","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":87002,"user_tz":240,"timestamp":1653425031155},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Mounted at /content/gdrive\n/content\n"}],"execution_count":1},{"cell_type":"code","source":"import pandas as pd\namazon_reviews = pd.read_csv('Reviews.csv')\n# Seleccionando solo los primeros 10,000 registros para calculo mas rapido\n#amazon_reviews = amazon_reviews[:10000]\namazon_reviews.head()","metadata":{"id":"zp73tMGp6haN","colab":{"height":337,"base_uri":"https://localhost:8080/"},"cell_id":"f7afa1b88d9143d2944af84b1d9735a3","outputId":"aa85f9ba-db9d-46e3-d875-e47102e3f291","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":8839,"user_tz":240,"timestamp":1653425118618},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":" Id ProductId UserId ProfileName \\\n0 1 B001E4KFG0 A3SGXH7AUHU8GW delmartian \n1 2 B00813GRG4 A1D87F6ZCVE5NK dll pa \n2 3 B000LQOCH0 ABXLMWJIXXAIN Natalia Corres \"Natalia Corres\" \n3 4 B000UA0QIQ A395BORC6FGVXV Karl \n4 5 B006K2ZZ7K A1UQRSCLF8GW1T Michael D. Bigham \"M. Wassir\" \n\n HelpfulnessNumerator HelpfulnessDenominator Score Time \\\n0 1 1 5 1303862400 \n1 0 0 1 1346976000 \n2 1 1 4 1219017600 \n3 3 3 2 1307923200 \n4 0 0 5 1350777600 \n\n Summary Text \n0 Good Quality Dog Food I have bought several of the Vitality canned d... \n1 Not as Advertised Product arrived labeled as Jumbo Salted Peanut... \n2 \"Delight\" says it all This is a confection that has been around a fe... \n3 Cough Medicine If you are looking for the secret ingredient i... \n4 Great taffy Great taffy at a great price. There was a wid... ","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
IdProductIdUserIdProfileNameHelpfulnessNumeratorHelpfulnessDenominatorScoreTimeSummaryText
01B001E4KFG0A3SGXH7AUHU8GWdelmartian1151303862400Good Quality Dog FoodI have bought several of the Vitality canned d...
12B00813GRG4A1D87F6ZCVE5NKdll pa0011346976000Not as AdvertisedProduct arrived labeled as Jumbo Salted Peanut...
23B000LQOCH0ABXLMWJIXXAINNatalia Corres \"Natalia Corres\"1141219017600\"Delight\" says it allThis is a confection that has been around a fe...
34B000UA0QIQA395BORC6FGVXVKarl3321307923200Cough MedicineIf you are looking for the secret ingredient i...
45B006K2ZZ7KA1UQRSCLF8GW1TMichael D. Bigham \"M. Wassir\"0051350777600Great taffyGreat taffy at a great price. There was a wid...
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":3}],"execution_count":3},{"cell_type":"code","source":"amazon_reviews.shape","metadata":{"id":"zJSLM18R0Ytm","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"ef87198733934c188d92b15c10c15ba0","outputId":"d96039e1-e30a-4f57-8d2d-eed5afec4f1c","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":217,"user_tz":240,"timestamp":1653425152489},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"(568454, 10)"},"metadata":{},"execution_count":4}],"execution_count":4},{"cell_type":"markdown","source":"Columnas del dataset:\n\n- **ID**: valor unico para cada fila\n- **ProductId**: una referencia del producto sobre la reseña\n- **UserId**: una referencia del usuario que dejo el review\n- **HelpfulnessNumerator**: numero de lectores que han indicado que la reseña ayuda o es interesante\n- **HelpfulnessDenominator**: numero total de personas que han dado una indicacion de si la reseña ha sido util o no \n- **Score**: rating (1-5)\n- **Time**: fecha formato timestamp que indica cuando se creo la revision \n- **Summary**: El resumen escrito por el usuario de lo que trata la reseña\n- **Text**: la revisión escrita por el usuario","metadata":{"id":"IsuNvQ957m-W","cell_id":"d0b61332062740749e6f33ff047d642a","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Miremos la distribucion del numero de palabras por review","metadata":{"id":"dq6TigUN8Zxy","cell_id":"fad108dbcbe04f648329310851fb09a7","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"words_per_review = amazon_reviews.Text.apply(lambda x: len(x.split(\" \")))\nwords_per_review.hist(bins = 100)\nplt.title('Numero de palabras por revision')\nplt.xlabel('Palabras')\nplt.ylabel('Frecuencia')","metadata":{"id":"FRoXoAfh8cUJ","colab":{"height":312,"base_uri":"https://localhost:8080/"},"cell_id":"c066c110f5f34d0788fe3667d97fae93","outputId":"383dc4ee-a6ba-42c8-9d08-288c9c92e3e6","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":4415,"user_tz":240,"timestamp":1653425197636},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"Text(0, 0.5, 'Frecuencia')"},"metadata":{},"execution_count":7},{"output_type":"display_data","data":{"text/plain":"
","image/png":"iVBORw0KGgoAAAANSUhEUgAAAZgAAAEWCAYAAABbgYH9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3de5QdZZnv8e+PhJsgJIDTkyFoQKIeLhJJCxlHnEYUAl4CDuOAKEEukRGOegaPhKMDUWQtdAaZARQnSCSMSLiNJkuCGJHWccZAuMQkgJhOCJIYEiAh2IAwwef8Ue8O1Tt7d+/s9Lu7O/w+a+3VtZ96q+qp2jv9pN56u0oRgZmZWX/bbqATMDOzbZMLjJmZZeECY2ZmWbjAmJlZFi4wZmaWhQuMmZll4QJjr1mSxkgKScNbvN3TJP2ywbbTJH0vd05DmaQ3SuqWNKyPdqdI+kmr8jIXGOuDpBWS1krapRQ7U1LnAKZltklE/C4ido2IV/pod0NEHN2qvMwFxhozDPjsQCdRrdVnHoPdYD0ejeQ1WHO3reMCY434J+DzkkZUz6jVzSSpU9KZafo0Sf8l6XJJz0paLuldKf5EOjuaXFp2R0n/LOl3ktZI+rakndO8DkkrJZ0v6Ungu6n9v0j6fXr9i6Qda+2EpGFp3U9LWg58oGr+7pKulbRa0ipJX63X7ZK6rm6VdJOkP0h6QNIhpflTJS1L8x6WdEK9gyvpX9OxeE7S/ZKOqGqyUy/bWZGOxyLgeUnDe9u2pP0l/VzShnQcbqqTU+VznZKO62pJn6/6nGoe91qfU431l78XzwDT+vjsH5H0wdLywyU9JenQ6u9gWvfytP+PSTqlFP9laR3vkrQgHYsFkt5Vmtcp6eKU4x8k/UTSXvU+Q6vNBcYacR/QCXy+j3b1HA4sAvYEvg/MAt4J7A98HLhK0q6p7aXAW4Bxaf7ewIWldf05sAfwJmAK8EVgQmp/CHAY8KU6eZwFfBB4B9AOnFg1/zpgY9ruO4CjgTN72a9JwC0pn+8DP5S0fZq3DDgC2B34MvA9SaPqrGdByr+ynlsk7dTgdgBOpiiWIyJiYx/bvhj4CTASGA1c2cv+ARwJjKU4FudLel+K93Xcqz+nWg4HlgNtwCX0/tnfmPaz4hjg6Yh4oLxCFV25VwDHRsTrgXcBC6s3LGkP4PbUdk/gG8DtkvYsNfsY8Engz4AdaP77/9oVEX75VfcFrADeBxwEbADeQPFLtzPNHwMEMLy0TCdwZpo+DVhamndwat9Wij1D8UtFwPPAm0vz/hJ4LE13AC8DO5XmLwOOK70/BlhRZ19+Bpxden90JXeKX3IvATuX5p8M3F1nXdOA+aX32wGrgSPqtF8ITCodk1/2cszXA4c0sp30+Zzex2dY3vb1wHRgdB/LVD7Xt5ViXweu7eu41/qcaqz/NOB3pfd9ffb7A38AXpfe3wBcWP0dBHYBngX+pvxZVh934BPAvVXzfwWcVvoOf6k079PAjwf63+NQe/kMxhoSEUuAHwFTm1h8TWn6xbS+6tiuFMXrdcD9KrrTngV+nOIVT0XEH0vv/wJ4vPT+8RSr5S+AJ6raVrwJ2B5YXdr2v1H877WeTeuKiD8BKyvblnSqpIWldR0E1OxikfT51AW0IbXdvapt3e1Uz29g21+g+GV+r6SHJJ3ey/5Vr7t8bPs67tWfU1/r7vWzj4gu4BHgQ5JeB3yY4myuh4h4Hvg74GyKz/J2SW+rse3q/Cv7sHfp/ZOl6RcovqO2BVxgbEtcRNHNVP5H+Hz6+bpS7M+bXP/TFMXmwIgYkV67R0T5H3b17b9/T1EcKt6YYrWsBvapalvxBMUZzF6lbe8WEQf2ku+mdUnajqLL6feS3gRcA5wL7BkRI4AlFL/Ye0jXW74AfBQYmdpuqGpbczul+VGa3+u2I+LJiDgrIv4C+BTwLUn7N7KP9Dy2fR33Rm7TXm7TyGdf6SabBDycis7mK424MyLeD4wCfkNxPKpV51/Zh1UN5G0NcoGxhqV/0DcBnynFnqL4R/lxFRfRTwfe3OT6/0Txy+BySX8GIGlvScf0stiNwJckvSFdhL0QqPd3IzcDn5E0WtJISmdjEbGa4trEZZJ2k7SdpDdL+utetj1e0kfSxeXPURSo+RTdNAE8lfbhkxRnEbW8nuK6z1PAcEkXArs1uJ1aet22pL+VNDq9XZ/a/qmXffxHSa+TdCDF9YjKoIAtOe59avCzn0XRrfn31Dh7Scu0SZqUrsW8BHTX2b+5wFskfSwNGPg74ACKs3TrJy4wtqW+QvFLrOws4P9SXEs5EPjvrVj/+UAXMF/Sc8BPgbf20v6rFIMQFgGLgQdSrJZrgDuBX6d2/1E1/1SKi7kPU/zyvZXif8H1zKbojllP0af/kYj4n4h4GLiMok9/DcV1p/+qs447KbqCfkvRRfNHqrq86m2n1soa2PY7gXskdQNzgM9GxPJe9vHnFJ/HXcA/R0TlDxW35Lg3qtfPPv0n4FcUF+5rjn6j+J32DxRnKOuAv6YoSD1ExDMUAz7Oo/jefgH4YEQ8vZX7YCVKF7DMbAtImgbsHxEfH+hccpA0BngM2D6KkWlmW8xnMGZmloULjJmZZeEuMjMzy8JnMGZmloVvMJfstddeMWbMmKaWff7559lll+qBVYOTc81nKOXrXPN4LeZ6//33Px0Rb6g5c6BvJTBYXuPHj49m3X333U0v22rONZ+hlK9zzeO1mCtwX/hWMWZm1kouMGZmloULjJmZZeECY2ZmWbjAmJlZFi4wZmaWhQuMmZll4QJjZmZZuMCYmVkW2QqMpBmS1kpaUordlJ4VvlDSCkkLU3yMpBdL875dWma8pMWSuiRdIUkpvoekeZKWpp8jU1ypXZekRZIOzbWPtYyZevuml5nZa1nOM5jrgInlQET8XUSMi4hxwG30fKLgssq8iDi7FL+a4omJY9Orss6pwF0RMZbiaXuVx98eW2o7JS1vZmYtlq3ARMQvKB5Zupl0FvJRiud61yVpFLBbRMxP97y5Hjg+zZ4EzEzTM6vi16fb5MwHRqT1mJlZCw3UNZgjgDURsbQU21fSg5J+LumIFNsbWFlqszLFANqieEY3wJNAW2mZJ+osY2ZmLTJQt+s/mZ5nL6uBN0bEM5LGAz+UdGCjK4uIkLTFT06TNIWiG422tjY6Ozu3dBUAdHd3b1r2vINffXx5s+vLqZzrYDeUcoWhla9zzcO59tTyAiNpOPARYHwlFhEvAS+l6fslLQPeAqwCRpcWH51iAGskjYqI1akLbG2KrwL2qbNMDxExHZgO0N7eHh0dHU3tU2dnJ5VlTytd3F9xSnPry6mc62A3lHKFoZWvc83DufY0EF1k7wN+ExGbur4kvUHSsDS9H8UF+uWpC+w5SRPSdZtTgdlpsTnA5DQ9uSp+ahpNNgHYUOpKy2Lxqg0eOWZmViXnMOUbgV8Bb5W0UtIZadZJbH5x/z3AojRs+Vbg7IioDBD4NPAdoAtYBtyR4pcC75e0lKJoXZric4Hlqf01aXkzM2uxbF1kEXFynfhpNWK3UQxbrtX+PuCgGvFngKNqxAM4ZwvTNTOzfua/5DczsyxcYMzMLAsXGDMzy8IFxszMsnCBMTOzLFxgzMwsCxcYMzPLwgXGzMyycIExM7MsXGDMzCwLFxgzM8vCBcbMzLJwgTEzsyxcYMzMLAsXGDMzy8IFxszMsnCBMTOzLFxgzMwsCxcYMzPLwgXGzMyyyFZgJM2QtFbSklJsmqRVkham13GleRdI6pL0qKRjSvGJKdYlaWopvq+ke1L8Jkk7pPiO6X1Xmj8m1z6amVl9Oc9grgMm1ohfHhHj0msugKQDgJOAA9My35I0TNIw4JvAscABwMmpLcDX0rr2B9YDZ6T4GcD6FL88tTMzsxbLVmAi4hfAugabTwJmRcRLEfEY0AUcll5dEbE8Il4GZgGTJAl4L3BrWn4mcHxpXTPT9K3AUam9mZm10EBcgzlX0qLUhTYyxfYGnii1WZli9eJ7As9GxMaqeI91pfkbUnszM2uh4S3e3tXAxUCkn5cBp7c4h00kTQGmALS1tdHZ2dnUetp2hvMO3rhZvNn15dTd3T0o86plKOUKQytf55qHc+2ppQUmItZUpiVdA/wovV0F7FNqOjrFqBN/BhghaXg6Sym3r6xrpaThwO6pfa18pgPTAdrb26Ojo6Op/bryhtlctnjzQ7nilObWl1NnZyfN7merDaVcYWjl61zzcK49tbSLTNKo0tsTgMoIsznASWkE2L7AWOBeYAEwNo0Y24FiIMCciAjgbuDEtPxkYHZpXZPT9InAz1J7MzNroWxnMJJuBDqAvSStBC4COiSNo+giWwF8CiAiHpJ0M/AwsBE4JyJeSes5F7gTGAbMiIiH0ibOB2ZJ+irwIHBtil8L/LukLopBBifl2kczM6svW4GJiJNrhK+tEau0vwS4pEZ8LjC3Rnw5xSiz6vgfgb/domTNzKzf+S/5zcwsCxcYMzPLwgXGzMyycIExM7MsXGDMzCwLFxgzM8vCBcbMzLJwgTEzsyxcYMzMLAsXGDMzy8IFxszMsnCBMTOzLFxgzMwsCxcYMzPLwgXGzMyycIExM7MsXGDMzCwLFxgzM8vCBcbMzLJwgTEzsyyyFRhJMyStlbSkFPsnSb+RtEjSDySNSPExkl6UtDC9vl1aZrykxZK6JF0hSSm+h6R5kpamnyNTXKldV9rOobn20czM6st5BnMdMLEqNg84KCLeDvwWuKA0b1lEjEuvs0vxq4GzgLHpVVnnVOCuiBgL3JXeAxxbajslLW9mZi2WrcBExC+AdVWxn0TExvR2PjC6t3VIGgXsFhHzIyKA64Hj0+xJwMw0PbMqfn0U5gMj0nrMzKyFBvIazOnAHaX3+0p6UNLPJR2RYnsDK0ttVqYYQFtErE7TTwJtpWWeqLOMmZm1yPCB2KikLwIbgRtSaDXwxoh4RtJ44IeSDmx0fRERkqKJPKZQdKPR1tZGZ2fnlq4CgLad4byDN24Wb3Z9OXV3dw/KvGoZSrnC0MrXuebhXHtqeYGRdBrwQeCo1O1FRLwEvJSm75e0DHgLsIqe3WijUwxgjaRREbE6dYGtTfFVwD51lukhIqYD0wHa29ujo6OjqX268obZXLZ480O54pTm1pdTZ2cnze5nqw2lXGFo5etc83CuPbW0i0zSROALwIcj4oVS/A2ShqXp/Sgu0C9PXWDPSZqQRo+dCsxOi80BJqfpyVXxU9NosgnAhlJXmpmZtUi2MxhJNwIdwF6SVgIXUYwa2xGYl0Ybz08jxt4DfEXS/wB/As6OiMoAgU9TjEjbmeKaTeW6zaXAzZLOAB4HPpric4HjgC7gBeCTufbRzMzqy1ZgIuLkGuFr67S9Dbitzrz7gINqxJ8BjqoRD+CcLUrWzMz6nf+S38zMsnCBMTOzLFxgzMwsCxcYMzPLwgXGzMyycIExM7MsXGDMzCwLFxgzM8uioQKTbtWyQFK3pJclvSLpudzJmZnZ0NXoGcxVwMnAUopbtpwJfDNXUmZmNvQ13EUWEV3AsIh4JSK+y+ZPqzQzM9uk0XuRvSBpB2ChpK9TPL/F12/MzKyuRovEJ4BhwLnA8xTPW/mbXEmZmdnQ19AZTEQ8niZfBL6cLx0zM9tW9FpgJN0cER+VtBjY7JHEEfH2bJmZmdmQ1tcZzGfTzw/mTsTMzLYtvRaY0qOGtwNWR8QfASTtDLRlzs3MzIawRi/y30LxKOOKV1LMzMyspkYLzPCIeLnyJk3vkCclMzPbFjRaYJ6S9OHKG0mTgKfzpGRmZtuCRv/Q8mzgBklXAQKeAE7NlpWZmQ15DZ3BRMSyiJgAHAD8r4h4V7p1TK8kzZC0VtKSUmwPSfMkLU0/R6a4JF0hqUvSIkmHlpaZnNovlTS5FB8vaXFa5gpJ6m0bZmbWOo3eTXlHSR8DPgP8g6QLJV3YwKLXsfk9y6YCd0XEWOCu9B7gWGBsek0Brk7b3gO4CDgcOAy4qFQwrgbOKi03sY9tmJlZizR6DWY2MAnYSHGrmMqrVxHxC2BdVXgSMDNNzwSOL8Wvj8J8YISkUcAxwLyIWBcR64F5wMQ0b7eImB8RAVxfta5a2zAzsxZp9BrM6Ijor7snt5X+vuZJXv17mr0pru1UrEyx3uIra8R720YPkqZQnC3R1tZGZ2dnE7sDbTvDeQdv3Cze7Ppy6u7uHpR51TKUcoWhla9zzcO59tRogflvSQdHxOL+3HhEhKTNbkHTqm1ExHRgOkB7e3t0dHQ0tY0rb5jNZYs3P5QrTmlufTl1dnbS7H622lDKFYZWvs41D+faU6NdZO8G7pf0aLoAv1jSoia3uSZ1b5F+rk3xVRR3aa4YnWK9xUfXiPe2DTMza5FGC0zlAvzRwIco7k32oSa3OQeojASbTHF9pxI/NY0mmwBsSN1cdwJHSxqZLu4fDdyZ5j2XHucsimHTs/vYhpmZtUijw5QfpziLeG+afqGRZSXdCPwKeKuklZLOAC4F3i9pKfC+9B5gLrAc6AKuAT6dtr0OuBhYkF5fSTFSm++kZZYBd6R4vW2YmVmLNHQNRtJFQDvwVuC7wPbA94C/6m25iDi5zqyjarQN4Jw665kBzKgRvw84qEb8mVrbMDOz1mm0i+wE4MOkockR8Xvg9bmSMjOzoa/RAvNyOsMIAEm75EvJzMy2BY0OU75Z0r9R/PHjWcDpFNdJrBdjpt6+aXrFpR8YwEzMzFqvoQITEf8s6f3AcxTXYS6MiHlZMzMzsyGt0TMYUkFxUTEzs4Y0OorsD6TrLxQPGtseeD4idsuVmJmZDW2NdpFtGjGW/qhxEjAhV1JmZjb0NTqKbJN0t+MfUtzl2MzMrKZGu8g+Unq7HcUfXf4xS0ZmZrZNaPQif/m+YxuBFRTdZGZmZjU1eg3mk7kTMTOzbUujj0yeKWlE6f1ISZvdG8zMzKyi0Yv8b4+IZytv0qOL35EnJTMz2xY0WmC2S89iAUDSHmzBH2mamdlrT6NF4jLgV5JuSe//FrgkT0pmZrYtaPQi//WS7gPem0IfiYiH86VlZmZD3Zb8oeUeFLeHuQp4StK+mXIyM7NtQKOjyC4CzgcuSKHKEy3NzMxq8hMtzcwsCz/R0szMsmi0wFQ/0fKnNPlES0lvlbSw9HpO0uckTZO0qhQ/rrTMBZK6JD0q6ZhSfGKKdUmaWorvK+meFL9J0g7N5GpmZs3rs8Ck2/PfBNwK3MarT7S8spkNRsSjETEuIsYB44EXgB+k2ZdX5kXE3LT9A4CTgAOBicC3JA2TNAz4JnAscABwcmoL8LW0rv2B9cAZzeRqZmbN63OYckSEpLkRcTD9/0TLo4BlEfF4UcdqmgTMioiXgMckdQGHpXldEbEcQNIsYJKkRyiGU38stZkJTAOu7ufczcysF43+oeUDkt4ZEQv6efsnATeW3p8r6VTgPuC8dEuavYH5pTYrUwzgiar44cCewLMRsbFG+x4kTQGmALS1tdHZ2dnUTrTtDOcdvLHXNs2uu791d3cPmlz6MpRyhaGVr3PNw7n21GiBORz4uKQVFCPJRHFy8/ZmN5yui3yYV4c+Xw1cTDGQ4GKKuwec3uz6GxER04HpAO3t7dHR0dHUeq68YTaXLe79UK44pbl197fOzk6a3c9WG0q5wtDK17nm4Vx76vW3oqQ3RsTvyPP0ymOBByJiDUDlZ9ruNcCP0ttVwD6l5UanGHXiz1AMRhiezmLK7c3MrEX6usj/Q4CIeBz4RkQ8Xn5t5bZPptQ9JmlUad4JwJI0PQc4SdKO6e4BY4F7gQXA2DRibAeK7rY5aTj13cCJafnJwOytzNXMzLZQX11k5Svv+/XXRtPf0bwf+FQp/HVJ4yi6yFZU5kXEQ5JuBh6meJrmORHxSlrPucCdwDBgRkQ8lNZ1PjBL0leBB4Fr+yt3MzNrTF8FJupMb5WIeJ7iYnw59ole2l9Cjbs3p6HMc2vEl/PqSDMzMxsAfRWYQyQ9R3Ems3Oahlcv8u+WNTszMxuyei0wETGsVYmYmdm2ZUtu129mZtYwFxgzM8vCBcbMzLJwgTEzsyxcYMzMLAsXGDMzy8IFxszMsnCBMTOzLFxgzMwsCxcYMzPLwgXGzMyycIExM7MsXGDMzCwLFxgzM8vCBcbMzLJwgTEzsyxcYMzMLAsXGDMzy2LACoykFZIWS1oo6b4U20PSPElL08+RKS5JV0jqkrRI0qGl9UxO7ZdKmlyKj0/r70rLqvV7aWb22jXQZzBHRsS4iGhP76cCd0XEWOCu9B7gWGBsek0BroaiIAEXAYcDhwEXVYpSanNWabmJ+XfHzMwqBrrAVJsEzEzTM4HjS/HrozAfGCFpFHAMMC8i1kXEemAeMDHN2y0i5kdEANeX1mVmZi2g4vfvAGxYegxYDwTwbxExXdKzETEizRewPiJGSPoRcGlE/DLNuws4H+gAdoqIr6b4PwIvAp2p/ftS/Ajg/Ij4YFUOUyjOiGhraxs/a9aspvZl7boNrHmx8fYH7717U9vpD93d3ey6664Dtv0tMZRyhaGVr3PN47WY65FHHnl/qReqh+FbvfbmvTsiVkn6M2CepN+UZ0ZESMpa/SJiOjAdoL29PTo6Oppaz5U3zOayxY0fyhWnNLed/tDZ2Umz+9lqQylXGFr5Otc8nGtPA9ZFFhGr0s+1wA8orqGsSd1bpJ9rU/NVwD6lxUenWG/x0TXiZmbWIgNSYCTtIun1lWngaGAJMAeojASbDMxO03OAU9NosgnAhohYDdwJHC1pZLq4fzRwZ5r3nKQJqavt1NK6zMysBQaqi6wN+EEaOTwc+H5E/FjSAuBmSWcAjwMfTe3nAscBXcALwCcBImKdpIuBBandVyJiXZr+NHAdsDNwR3qZmVmLDEiBiYjlwCE14s8AR9WIB3BOnXXNAGbUiN8HHLTVyZqZWVMG2zBlMzPbRrjAmJlZFi4wZmaWhQuMmZll4QJjZmZZuMCYmVkWLjBmZpaFC4yZmWXhAmNmZlm4wJiZWRYuMGZmloULjJmZZeECY2ZmWbjAmJlZFi4wZmaWhQuMmZll4QJjZmZZuMCYmVkWLjBmZpbF8IFO4LVozNTbe7xfcekHBigTM7N8Wn4GI2kfSXdLeljSQ5I+m+LTJK2StDC9jistc4GkLkmPSjqmFJ+YYl2Sppbi+0q6J8VvkrRDa/fSzMwGootsI3BeRBwATADOkXRAmnd5RIxLr7kAad5JwIHAROBbkoZJGgZ8EzgWOAA4ubSer6V17Q+sB85o1c6ZmVmh5QUmIlZHxANp+g/AI8DevSwyCZgVES9FxGNAF3BYenVFxPKIeBmYBUySJOC9wK1p+ZnA8Xn2xszM6lFEDNzGpTHAL4CDgH8ATgOeA+6jOMtZL+kqYH5EfC8tcy1wR1rFxIg4M8U/ARwOTEvt90/xfYA7IuKgGtufAkwBaGtrGz9r1qym9mPtug2sebGpRQE4eO/dm194C3V3d7Prrru2bHtbYyjlCkMrX+eax2sx1yOPPPL+iGivNW/ALvJL2hW4DfhcRDwn6WrgYiDSz8uA03PmEBHTgekA7e3t0dHR0dR6rrxhNpctbv5Qrjilue02o7Ozk2b3s9WGUq4wtPJ1rnk4154GpMBI2p6iuNwQEf8BEBFrSvOvAX6U3q4C9iktPjrFqBN/BhghaXhEbKxqb2ZmLTIQo8gEXAs8EhHfKMVHlZqdACxJ03OAkyTtKGlfYCxwL7AAGJtGjO1AMRBgThR9fncDJ6blJwOzc+6TmZltbiDOYP4K+ASwWNLCFPt/FKPAxlF0ka0APgUQEQ9Juhl4mGIE2jkR8QqApHOBO4FhwIyIeCit73xglqSvAg9SFDQzM2uhlheYiPgloBqz5vayzCXAJTXic2stFxHLKUaZmZnZAPGtYszMLAsXGDMzy8IFxszMsnCBMTOzLHw35UGgfHdl31nZzLYVPoMxM7MsXGDMzCwLFxgzM8vCBcbMzLJwgTEzsyxcYMzMLAsPUx5kPGTZzLYVPoMxM7MsXGDMzCwLFxgzM8vCBcbMzLLwRf5BzBf8zWwo8xmMmZll4TOYIcJnM2Y21PgMxszMsthmz2AkTQT+FRgGfCciLh3glPqNz2bMbCjYJguMpGHAN4H3AyuBBZLmRMTDA5tZ/ysXmzIXHjMbaNtkgQEOA7oiYjmApFnAJGCbKzD11Cs85x28kdPqzKtwcTKz/rCtFpi9gSdK71cCh1c3kjQFmJLedkt6tMnt7QU83eSyLfWZBnLV11qUTN+GzHFNhlK+zjWP12Kub6o3Y1stMA2JiOnA9K1dj6T7IqK9H1LKzrnmM5Tyda55ONeettVRZKuAfUrvR6eYmZm1yLZaYBYAYyXtK2kH4CRgzgDnZGb2mrJNdpFFxEZJ5wJ3UgxTnhERD2Xc5FZ3s7WQc81nKOXrXPNwriWKiNzbMDOz16BttYvMzMwGmAuMmZll4QKzlSRNlPSopC5JUwc6HwBJKyQtlrRQ0n0ptoekeZKWpp8jU1ySrkj5L5J0aObcZkhaK2lJKbbFuUmanNovlTS5hblOk7QqHduFko4rzbsg5fqopGNK8ezfEUn7SLpb0sOSHpL02RQfdMe2l1wH3bGVtJOkeyX9OuX65RTfV9I9abs3pcFESNoxve9K88f0tQ8tyvc6SY+Vju24FM/7PYgIv5p8UQwgWAbsB+wA/Bo4YBDktQLYqyr2dWBqmp4KfC1NHwfcAQiYANyTObf3AIcCS5rNDdgDWJ5+jkzTI1uU6zTg8zXaHpA+/x2BfdP3YlirviPAKODQNP164Lcpp0F3bHvJddAd23R8dk3T2wP3pON1M3BSin8b+Ps0/Wng22n6JOCm3vYhw/egXr7XASfWaJ/1e+AzmK2z6ZY0EfEyULklzWA0CZiZpmcCx5fi10dhPjBC0qhcSUTEL4B1W5nbMcC8iFgXEeuBecDEFuVazyRgVkS8FBGPAV0U34+WfEciYnVEPJCm/wA8QnFHi0F3bHvJtZ4BO7bp+HSnt9unVwDvBW5N8erjWjnetwJHSVIv+9Cvesm3nqzfAxeYrVPrljS9/UNplQB+Iul+FbfDAWiLiNVp+kmgLc+G0N8AAARpSURBVE0Phn3Y0twGOudzU3fCjEqXUy85tTzX1C3zDor/vQ7qY1uVKwzCYytpmKSFwFqKX7TLgGcjYmON7W7KKc3fAOzZqlxr5RsRlWN7STq2l0vasTrfqrz6JV8XmG3TuyPiUOBY4BxJ7ynPjOIceFCOTx/MuSVXA28GxgGrgcsGNp2eJO0K3AZ8LiKeK88bbMe2Rq6D8thGxCsRMY7ijiCHAW8b4JR6VZ2vpIOACyjyfidFt9f5rcjFBWbrDMpb0kTEqvRzLfADin8UaypdX+nn2tR8MOzDluY2YDlHxJr0D/hPwDW82s0x4LlK2p7iF/YNEfEfKTwoj22tXAfzsU35PQvcDfwlRVdS5Q/Vy9vdlFOavzvwTKtzrcp3YuqWjIh4CfguLTq2LjBbZ9DdkkbSLpJeX5kGjgaWpLwqI0EmA7PT9Bzg1DSaZAKwodSl0ipbmtudwNGSRqZulKNTLLuq61MnUBzbSq4npVFE+wJjgXtp0Xck9fNfCzwSEd8ozRp0x7ZeroPx2Ep6g6QRaXpnimdMPULxi/vE1Kz6uFaO94nAz9KZY7196Fd18v1N6T8ZorheVD62+b4HWzoqwK+aozB+S9Ev+8VBkM9+FKNVfg08VMmJoh/4LmAp8FNgjxQXxcPZlgGLgfbM+d1I0f3xPxT9umc0kxtwOsWF0i7gky3M9d9TLovSP85RpfZfTLk+Chzbyu8I8G6K7q9FwML0Om4wHttech10xxZ4O/BgymkJcGHp39m96RjdAuyY4jul911p/n597UOL8v1ZOrZLgO/x6kizrN8D3yrGzMyycBeZmZll4QJjZmZZuMCYmVkWLjBmZpaFC4yZmWXhAmPWzyS9ku5Yu0TSLZJe10vb0yRd1cf6pkn6fP9napaXC4xZ/3sxIsZFxEHAy8DZuTdY+qtys0HDBcYsr/8E9pf0IRXPB3lQ0k8ltVU37KPNIZJ+lZ7NcVZq3yHpPyXNAR5OsR+mm5w+VLnRabr54XXpjGqxpP/Tgv02w//rMcsknVUcC/wY+CUwISJC0pnAF4Dzqhbprc3bKZ7XsQvwoKTbU/xQ4KAobgEPcHpErEu3CVkg6TZgDLB3OqOicisRs9xcYMz6387pdulQnMFcC7wVuCndE2oH4LEay43upc3siHgReFHS3RQ3K3wWuLdUXAA+I+mENL0PxT2vHgX2k3QlcDvwk/7YSbO+uIvMrP9VrsGMi4j/HcXDsK4EroqIg4FPUdyzqlpvbarv6VR5/3wlIKkDeB/wlxFxCMU9qXaK4oFRhwCdFNeDvrOV+2fWEBcYs9bYnVdvd17v+ea9tZmk4nnrewIdFHcSrrX8+oh4QdLbKLrUkLQXsF1E3AZ8iaJbzSw7Fxiz1pgG3CLpfuDpJtosorhF/Hzg4oj4fY3lfwwMl/QIcGlqC8WTCDtTt933KB4+ZZad76ZsZmZZ+AzGzMyycIExM7MsXGDMzCwLFxgzM8vCBcbMzLJwgTEzsyxcYMzMLIv/DyBANFniixocAAAAAElFTkSuQmCC\n"},"metadata":{"needs_background":"light"}}],"execution_count":7},{"cell_type":"code","source":"words_per_review.mean()","metadata":{"id":"7mcYXdhd8f1Z","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"4052dac4840b4499a47632487e1b3171","outputId":"d9c2e6ec-4c61-4e9d-fc56-a78f25339f7c","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":226,"user_tz":240,"timestamp":1653425215006},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"82.00552199474363"},"metadata":{},"execution_count":8}],"execution_count":8},{"cell_type":"markdown","source":"Distrobucion de los ratings","metadata":{"id":"rRCyuNQ38gmB","cell_id":"2001e82922244900afe3c0aab5c14eb4","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"amazon_reviews.Score.value_counts()","metadata":{"id":"CIio-PLz8iHJ","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"f6a7fb776db34abaa1103c8d48c5ce15","outputId":"50333a0b-a863-435d-d704-08ac133c0b5c","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":218,"user_tz":240,"timestamp":1653425218506},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"5 363122\n4 80655\n1 52268\n3 42640\n2 29769\nName: Score, dtype: int64"},"metadata":{},"execution_count":9}],"execution_count":9},{"cell_type":"code","source":"percent_val = 100 * amazon_reviews.Score.value_counts()/amazon_reviews.shape[0]\npercent_val","metadata":{"id":"wWAVP9ad8jSL","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"124af6c91ea34c00ae3e8bb40d00c7a3","outputId":"6b87f690-8603-4e97-96ad-a9c42eb561ff","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":215,"user_tz":240,"timestamp":1653425224397},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"5 63.878871\n4 14.188483\n1 9.194763\n3 7.501047\n2 5.236835\nName: Score, dtype: float64"},"metadata":{},"execution_count":10}],"execution_count":10},{"cell_type":"code","source":"percent_val.plot.bar()\nplt.title('Revisiones por scores')\nplt.xlabel('Score')\nplt.ylabel('Porcentaje (%)')","metadata":{"id":"GUlRhwhj8lEa","colab":{"height":309,"base_uri":"https://localhost:8080/"},"cell_id":"2dddfb71d53d40fbb42d721ab890b24a","outputId":"02a602e1-029b-43c2-82ce-8b01f03e8f19","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":636,"user_tz":240,"timestamp":1653425236417},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"Text(0, 0.5, 'Porcentaje (%)')"},"metadata":{},"execution_count":11},{"output_type":"display_data","data":{"text/plain":"
","image/png":"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\n"},"metadata":{"needs_background":"light"}}],"execution_count":11},{"cell_type":"markdown","source":"la distribucion es asimetrica, con un gran valor para los ratings de 5s y pocos par 3,2 y 1","metadata":{"id":"jJMFaoaW8oE7","cell_id":"1b27c5e9de4a47ef8072ed7c48ff9571","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Wordcloud","metadata":{"id":"z6JfoGI18uT2","cell_id":"279ac92dab214777b8387bef88a01ffa","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"word_cloud_text = ''.join(amazon_reviews.Text)\nprint(len(word_cloud_text))\n\nwordcloud = WordCloud(\n max_font_size=100,\n max_words=100,\n background_color=\"white\",\n scale=10,\n width=800,\n height=400\n).generate(word_cloud_text)\n\nplt.figure(figsize=(12,6))\nplt.imshow(wordcloud, interpolation=\"bilinear\")\nplt.axis(\"off\")\nplt.show()","metadata":{"id":"p4T2z4H-8vat","colab":{"height":374,"base_uri":"https://localhost:8080/"},"cell_id":"4cadd105dba544c99ff4924e4c8c358b","outputId":"ae7d7034-a376-4fcf-a86f-9cdc1c9a6a14","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":126653,"user_tz":240,"timestamp":1653425441082},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"247972188\n"},{"output_type":"display_data","data":{"text/plain":"
","image/png":"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\n"},"metadata":{"needs_background":"light"}}],"execution_count":12},{"cell_type":"markdown","source":"El wordcloud muestra que hat muchos reviews que hablan sobre temas de comida (cafe, sabores, sabor, bebidas), tambien se observan otras palabras como bueno, amor y el mejor","metadata":{"id":"qQK8Ykur80AZ","cell_id":"4e0ca5dc8f464a3cb18629533dacc2f4","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Estandarizacion de ratings para analisis de sentimiento","metadata":{"id":"oCIjwyvX9DnC","cell_id":"1420a9eb1e034610800e705d389e2531","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Para propositos del analisis de sentimiento convertiremos todos los ratings en valores binarios con las siguientes reglas:\n- ratings de 4 o 5 seran convertidos a 1 (positivo)\n- ratings de 1 o 2 seran convertidos a 0 (negativo)\n- ratings de 3 seran removidos del analisis","metadata":{"id":"CFpvV3IP9HPt","cell_id":"a39d97237bf547018738580e29752291","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"amazon_reviews['Sentiment_rating'] = np.where(amazon_reviews.Score > 3, 1, 0)\namazon_reviews['Sentiment_rating'].value_counts()","metadata":{"id":"XE3s2uHi9W0t","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"0b36956a6a99483bbdc89ba85f073fc5","outputId":"374ce431-7b43-47ad-ce0e-aa252ab4a443","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":230,"user_tz":240,"timestamp":1653425658306},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"1 443777\n0 124677\nName: Sentiment_rating, dtype: int64"},"metadata":{},"execution_count":13}],"execution_count":13},{"cell_type":"code","source":"# removiendo neutrales\namazon_reviews = amazon_reviews[amazon_reviews.Score != 3]","metadata":{"id":"Ne9ciavE9Gp-","cell_id":"ee06e9db81cf4d1ca7e78e5f8b17ce4a","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":218,"user_tz":240,"timestamp":1653425668029},"deepnote_cell_type":"code"},"outputs":[],"execution_count":14},{"cell_type":"code","source":"#rcParams['figure.figsize'] = 8, 5\namazon_reviews.Sentiment_rating.value_counts().plot.bar()\nplt.title('Scores luego de estandarizacion')\nplt.xlabel('Score')\nplt.ylabel('Frecuencia')","metadata":{"id":"7EFLK3p69ajb","colab":{"height":309,"base_uri":"https://localhost:8080/"},"cell_id":"664dd4402100416fb7f2dc2ce4da1ba9","outputId":"4756d0bc-3398-4ab0-a469-fa5ed8a0a356","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":731,"user_tz":240,"timestamp":1653425680669},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"Text(0, 0.5, 'Frecuencia')"},"metadata":{},"execution_count":16},{"output_type":"display_data","data":{"text/plain":"
","image/png":"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\n"},"metadata":{"needs_background":"light"}}],"execution_count":16},{"cell_type":"markdown","source":"## voy por aqui!!!","metadata":{"id":"1WI1bJFu3St2","cell_id":"73ab69ea8dd541df83f094c15fb84a8c","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Preprocesamiento","metadata":{"id":"yQHkuA0Y9e2q","cell_id":"b0c685ca52bf42c0b56bfe728ff3f5c2","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Recordemos que el preprocesamiento de texto y la normalizacion es crucial antes de desarrollar un modelo de NLP, algunos pasos importantes son:\n1. Convertir palabras a minusculas\n2. remover caracteres especiales\n3. remover stopwords y palabras de alta frecuencia\n4. Stemming y lemantizacion\n\nProcedamos con la primera fase","metadata":{"id":"DqxWLz1PAfer","cell_id":"3e7c82c1751248ce950bd8fbd8110210","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"amazon_reviews['reviews_text_new'] = amazon_reviews.Text.apply(lambda x: x.lower())","metadata":{"id":"nqPMAWzaAzWk","cell_id":"32d07a0d9e19489bb77900e8c1337113","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Las siguientes fases:","metadata":{"id":"417Vf1C4A1oI","cell_id":"33f1d1e3e8f94d25a144df14916d57a2","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"from nltk import word_tokenize\nimport nltk\n\nnltk.download('punkt')\n\ntoken_lists = [word_tokenize(each) for each in amazon_reviews.Text]\ntokens = [item for sublist in token_lists for item in sublist]\nprint(\"Numero de tokens unicos antes: \", len(set(tokens)))\n\ntoken_lists_lower = [word_tokenize(each) for each in amazon_reviews.reviews_text_new]\ntokens_lower = [item for sublist in token_lists_lower for item in sublist]\nprint(\"Numero de tokens unicos nuevos: \", len(set(tokens_lower)))","metadata":{"id":"zC_-ZELYA3Iz","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"434ffeb8665143a78b5497cdbdf5f1b2","outputId":"a462dd5e-1d0e-4e8c-e9f5-f2ef5843ec86","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":13629,"user_tz":240,"timestamp":1653305963072},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"[nltk_data] Downloading package punkt to /root/nltk_data...\n[nltk_data] Unzipping tokenizers/punkt.zip.\nNumero de tokens unicos antes: 27899\nNumero de tokens unicos nuevos: 22865\n"}],"execution_count":null},{"cell_type":"code","source":"(22865-27899)/27899","metadata":{"id":"2Ixj5AbQBFaN","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"c98f4396515f4086918d099070466134","outputId":"d001f26a-bb9d-4ed9-ee98-5d0435746927","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":22,"user_tz":240,"timestamp":1653305963073},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"-0.18043657478762679"},"metadata":{},"execution_count":18}],"execution_count":null},{"cell_type":"markdown","source":"El numero de tokens han bajado en cerca del 18% con la normalizacion","metadata":{"id":"EQBzVKWlA_-0","cell_id":"48c1f1e5ebca4926ba881750ffe7411b","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"¿Es la eliminación de caracteres especiales incluso una buena idea? ¿Cuáles son algunos ejemplos de caracteres que probablemente sería seguro eliminar y cuáles no?","metadata":{"id":"uIH8nmc8BMav","cell_id":"09785bbc4bd141aa88971ed39dae4d16","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Eliminar caracteres especiales es una decisión subjetiva, especialmente en casos como este. Las personas a menudo usan caracteres especiales para expresar sus emociones y pueden dejar una reseña como \"¡¡¡Este producto es el peor!!!\", mientras que una reseña positiva podría ser \"Este producto es el mejor\". ¡Me encantó!' Aquí, la presencia de signos de exclamación indica claramente algo sobre el sentimiento subyacente, por lo que eliminarlos puede no ser una buena idea.\n\nPor otro lado, eliminar la puntuación sin carga emocional, como las comas, los puntos y el punto y coma, probablemente sea seguro.\n\nEn aras de la simplicidad, procederemos eliminando todos los caracteres especiales; sin embargo, vale la pena tener en cuenta que esto es algo para revisar dependiendo de los resultados que obtengamos más adelante. Lo siguiente da una lista de todos los caracteres especiales en nuestro conjunto de datos:","metadata":{"id":"8SnPc7g6BV9N","cell_id":"d0b1b40852664b6e925e04a9a89c6720","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"# Seleccionando los caracteres no alfa numericos que no son espacios\nspecial_chars = amazon_reviews.reviews_text_new.apply(lambda x: [each for each in list(x) if not each.isalnum() and each != ' '])\n\n# obtener una lista de listas\nflat_list = [item for sublist in special_chars for item in sublist]\n\n# caracteres especiales unicos \nprint(set(flat_list))","metadata":{"id":"33IALa4i9gB8","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"ffb7825d044e4536ade821f37e1fe39c","outputId":"eddb3dd7-38e8-4a07-be73-205eb699d350","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":662,"user_tz":240,"timestamp":1653305969133},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"{')', '`', '§', '^', '~', \"'\", '[', ';', '.', '\"', '<', '{', '(', '#', '}', '®', '&', '@', '-', ':', '>', '=', '_', '!', '%', ',', '?', ']', '+', '$', '*', '/'}\n"}],"execution_count":null},{"cell_type":"markdown","source":"Ahora removamos los caracteres especiales de los reviews","metadata":{"id":"kJeTZCmfBrX8","cell_id":"fab80a93a4d74b0588c539e8ef2c9033","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"import re\nreview_backup = amazon_reviews.reviews_text_new.copy()\namazon_reviews.reviews_text_new = amazon_reviews.reviews_text_new.apply(\n lambda x: re.sub('[^A-Za-z0-9 ]+', ' ', x)\n)","metadata":{"id":"tI1Zh8MiBuZw","cell_id":"1ae200e61a074468af65bbe3d236a55b","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Miremos como se ven algunos de los resultados luego de remover esto","metadata":{"id":"btC5RULGBxLX","cell_id":"b32f9f1db48c4f19a13aca864c18edce","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"print(\"Review anterior:\")\nreview_backup.values[6]","metadata":{"id":"wITFs61CB0cx","colab":{"height":105,"base_uri":"https://localhost:8080/"},"cell_id":"bbb884e9f5084ca7973e0496db99f7e7","outputId":"e35044b9-2529-40b2-d67c-dbd297c43c4a","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":12,"user_tz":240,"timestamp":1653305972711},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Review anterior:\n"},{"output_type":"execute_result","data":{"text/plain":"\"this saltwater taffy had great flavors and was very soft and chewy. each candy was individually wrapped well. none of the candies were stuck together, which did happen in the expensive version, fralinger's. would highly recommend this candy! i served it at a beach-themed party and everyone loved it!\"","application/vnd.google.colaboratory.intrinsic+json":{"type":"string"}},"metadata":{},"execution_count":21}],"execution_count":null},{"cell_type":"code","source":"print(\"Review nuevo:\")\namazon_reviews.reviews_text_new[6]","metadata":{"id":"E8K8haC9B29Y","colab":{"height":105,"base_uri":"https://localhost:8080/"},"cell_id":"6adf9bbba0ca482ca959212323cfd2fb","outputId":"4c697f4b-7fe5-4038-a660-f7611313e536","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":494,"user_tz":240,"timestamp":1653305974276},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Review nuevo:\n"},{"output_type":"execute_result","data":{"text/plain":"'this saltwater taffy had great flavors and was very soft and chewy each candy was individually wrapped well none of the candies were stuck together which did happen in the expensive version fralinger s would highly recommend this candy i served it at a beach themed party and everyone loved it '","application/vnd.google.colaboratory.intrinsic+json":{"type":"string"}},"metadata":{},"execution_count":22}],"execution_count":null},{"cell_type":"markdown","source":"El numero de tokens unicos que se han borrado son","metadata":{"id":"viiMJcdXB9fA","cell_id":"cb64564df97742d5bdf22f6abbbf65f5","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"token_lists = [word_tokenize(each) for each in amazon_reviews.Text]\ntokens = [item for sublist in token_lists for item in sublist]\nprint(\"Numero de token unicos antes: \", len(set(tokens)))\n\ntoken_lists = [word_tokenize(each) for each in amazon_reviews.reviews_text_new]\ntokens = [item for sublist in token_lists for item in sublist]\nprint(\"Numero de tokens unicos despues: \", len(set(tokens)))","metadata":{"id":"HkwuWkWZB8oz","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"e3bc65094c5b44d7846d3258d9189480","outputId":"a494a9f9-6de7-404f-c5b8-af28dd218897","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":9742,"user_tz":240,"timestamp":1653305985492},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Numero de token unicos antes: 27899\nNumero de tokens unicos despues: 18039\n"}],"execution_count":null},{"cell_type":"markdown","source":"## stopwords y palabras de alta/baja frecuencia\n\nVamos a remover estas palabras","metadata":{"id":"-EKO6f24CIbg","cell_id":"3caa616d07cf4b56b96bdcf595479137","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"import nltk\nnltk.download('stopwords')\n\nnoise_words = []\nstopwords_corpus = nltk.corpus.stopwords\neng_stop_words = stopwords_corpus.words('english')\nnoise_words.extend(eng_stop_words)\nprint(len(noise_words))\nnoise_words","metadata":{"id":"Z6uhTzzVCPB2","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"8beed5d9bb6f4755a801c9949878e5cc","outputId":"a4b74b78-0ea3-448e-8360-0df805d17a4c","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":12,"user_tz":240,"timestamp":1653305985493},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"[nltk_data] Downloading package stopwords to /root/nltk_data...\n[nltk_data] Unzipping corpora/stopwords.zip.\n179\n"},{"output_type":"execute_result","data":{"text/plain":"['i',\n 'me',\n 'my',\n 'myself',\n 'we',\n 'our',\n 'ours',\n 'ourselves',\n 'you',\n \"you're\",\n \"you've\",\n \"you'll\",\n \"you'd\",\n 'your',\n 'yours',\n 'yourself',\n 'yourselves',\n 'he',\n 'him',\n 'his',\n 'himself',\n 'she',\n \"she's\",\n 'her',\n 'hers',\n 'herself',\n 'it',\n \"it's\",\n 'its',\n 'itself',\n 'they',\n 'them',\n 'their',\n 'theirs',\n 'themselves',\n 'what',\n 'which',\n 'who',\n 'whom',\n 'this',\n 'that',\n \"that'll\",\n 'these',\n 'those',\n 'am',\n 'is',\n 'are',\n 'was',\n 'were',\n 'be',\n 'been',\n 'being',\n 'have',\n 'has',\n 'had',\n 'having',\n 'do',\n 'does',\n 'did',\n 'doing',\n 'a',\n 'an',\n 'the',\n 'and',\n 'but',\n 'if',\n 'or',\n 'because',\n 'as',\n 'until',\n 'while',\n 'of',\n 'at',\n 'by',\n 'for',\n 'with',\n 'about',\n 'against',\n 'between',\n 'into',\n 'through',\n 'during',\n 'before',\n 'after',\n 'above',\n 'below',\n 'to',\n 'from',\n 'up',\n 'down',\n 'in',\n 'out',\n 'on',\n 'off',\n 'over',\n 'under',\n 'again',\n 'further',\n 'then',\n 'once',\n 'here',\n 'there',\n 'when',\n 'where',\n 'why',\n 'how',\n 'all',\n 'any',\n 'both',\n 'each',\n 'few',\n 'more',\n 'most',\n 'other',\n 'some',\n 'such',\n 'no',\n 'nor',\n 'not',\n 'only',\n 'own',\n 'same',\n 'so',\n 'than',\n 'too',\n 'very',\n 's',\n 't',\n 'can',\n 'will',\n 'just',\n 'don',\n \"don't\",\n 'should',\n \"should've\",\n 'now',\n 'd',\n 'll',\n 'm',\n 'o',\n 're',\n 've',\n 'y',\n 'ain',\n 'aren',\n \"aren't\",\n 'couldn',\n \"couldn't\",\n 'didn',\n \"didn't\",\n 'doesn',\n \"doesn't\",\n 'hadn',\n \"hadn't\",\n 'hasn',\n \"hasn't\",\n 'haven',\n \"haven't\",\n 'isn',\n \"isn't\",\n 'ma',\n 'mightn',\n \"mightn't\",\n 'mustn',\n \"mustn't\",\n 'needn',\n \"needn't\",\n 'shan',\n \"shan't\",\n 'shouldn',\n \"shouldn't\",\n 'wasn',\n \"wasn't\",\n 'weren',\n \"weren't\",\n 'won',\n \"won't\",\n 'wouldn',\n \"wouldn't\"]"},"metadata":{},"execution_count":24}],"execution_count":null},{"cell_type":"markdown","source":"Encontremos las palabras de alta y baja frecuencia, que definiremos como el 1 % de las palabras que aparecen con más frecuencia en las reseñas, así como el 1 % de las palabras que aparecen con menos frecuencia en las reseñas (después de ajustar por mayúsculas y minúsculas y caracteres especiales).","metadata":{"id":"oonZd40DCRnt","cell_id":"f982d1c73dba40f6a59623f0e50ed612","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"one_percentile = int(len(set(tokens)) * 0.01)\ntop_1_percentile = Counter(tokens).most_common(one_percentile)\ntop_1_percentile[:10]","metadata":{"id":"phySUXSpCb7N","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"cadffbb65ec44b849dbb24b87317337f","outputId":"2525fa21-3767-4d51-d9a2-407d55de420a","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":264,"user_tz":240,"timestamp":1653305995000},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"[('the', 28122),\n ('i', 25705),\n ('and', 19980),\n ('a', 18505),\n ('it', 16143),\n ('to', 15137),\n ('of', 12067),\n ('is', 11063),\n ('this', 10530),\n ('br', 9361)]"},"metadata":{},"execution_count":25}],"execution_count":null},{"cell_type":"code","source":"pd.DataFrame(top_1_percentile[:10], columns=['Palabras','Frecuencia']).set_index('Palabras').plot(kind='bar')\nplt.title('Percentil 1 de palabras mas frecuentes')\nplt.xlabel('Palabras')\nplt.ylabel('Frecuencia')","metadata":{"id":"cIogh--ct3CJ","colab":{"height":376,"base_uri":"https://localhost:8080/"},"cell_id":"f925313e71894e2bae2f0ebc5faf5cb2","outputId":"091a5833-1592-4e74-8ce8-869c308a64df","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":832,"user_tz":240,"timestamp":1653306148618},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"Text(0, 0.5, 'Frecuencia')"},"metadata":{},"execution_count":35},{"output_type":"display_data","data":{"text/plain":"
","image/png":"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\n"},"metadata":{"needs_background":"light"}}],"execution_count":null},{"cell_type":"code","source":"bottom_1_percentile = Counter(tokens).most_common()[-one_percentile:]\nbottom_1_percentile[:10]","metadata":{"id":"lnFGqSCTCdU5","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"05f4902d0f9140589a08ef637600c87e","outputId":"f13d499a-643d-4cad-800b-7a5487c00a37","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":470,"user_tz":240,"timestamp":1653306224182},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"[('pruchase', 1),\n ('slick', 1),\n ('cloured', 1),\n ('innocuous', 1),\n ('espensive', 1),\n ('marketer', 1),\n ('strofoam', 1),\n ('destroyers', 1),\n ('ruth', 1),\n ('gleaning', 1)]"},"metadata":{},"execution_count":36}],"execution_count":null},{"cell_type":"code","source":"noise_words.extend([word for word,val in top_1_percentile])\nnoise_words.extend([word for word,val in bottom_1_percentile])","metadata":{"id":"g-A1HtiWCfH2","cell_id":"7f29d3cfcb7a491db01b44306d1f9a36","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Las stopwords y las palabras de alta/baja frecuencia ahora se han agregado a noise_words, que se eliminarán de las revisiones antes de entrenar los modelos de aprendizaje automático. \n\nEs poco probable que las stopwords sean tan útiles, ya que esperamos que aparezcan con la misma frecuencia en las críticas positivas y negativas. Las palabras poco comunes pueden ser más significativas y, en teoría, podrían indicar el sentimiento de la revisión","metadata":{"id":"n51bDGhkCiQF","cell_id":"5fb1727b662f403891305c98cda30a5a","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Stemming y Lemantizacion\n\nSi quieren profundizar en los conceptos de [stemming](http://www.nltk.org/api/nltk.stem.html?highlight=lemmatizer), y[lemmatization](http://www.nltk.org/api/nltk.stem.html?highlight=lemmatizer#module-nltk.stem.wordnet) y otros tipos de normalizaciones pueden encontrar una buena introduccion en: https://nlp.stanford.edu/IR-book/html/htmledition/stemming-and-lemmatization-1.html.","metadata":{"id":"TFmioUxSCtCO","cell_id":"f1cf83798b40484781af1c485f333a89","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"from nltk.stem import PorterStemmer, WordNetLemmatizer, LancasterStemmer\n\nnltk.download('wordnet')\n\nfrom nltk.corpus import wordnet\n\nporter = PorterStemmer()\nlancaster = LancasterStemmer()\nlemmatizer = WordNetLemmatizer()","metadata":{"id":"uWkdbiEtC8Qc","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"066514759d8447e5b604238a97e4e8f1","outputId":"7d673d45-6f69-4e6c-b741-b6e1e33d734e","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":879,"user_tz":240,"timestamp":1653261196018},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"[nltk_data] Downloading package wordnet to /root/nltk_data...\n[nltk_data] Unzipping corpora/wordnet.zip.\n"}],"execution_count":null},{"cell_type":"markdown","source":"Los algoritmos de **Stemming** funcionan cortando el final o el principio de la palabra, teniendo en cuenta una lista de prefijos y sufijos comunes que se pueden encontrar.\n\nPor otro lado, la **lematización** toma en consideración el análisis morfológico de las palabras. Por lo tanto, la **lematización** tiene en cuenta la gramática de la palabra e intenta encontrar la palabra raíz en lugar de simplemente llegar a la palabra raíz mediante métodos de fuerza bruta.","metadata":{"id":"OWrUPKFfC9sa","cell_id":"1e8b9ad59445480eb33dc0ef51a82425","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"print(\"Lancaster Stemmer\")\nprint(lancaster.stem(\"trouble\"))\nprint(lancaster.stem(\"troubling\"))\nprint(lancaster.stem(\"troubled\"))\n\n# Proveer una palabra que sera lemantizada\nprint(\"WordNet Lemmatizer\")\nprint(lemmatizer.lemmatize(\"trouble\", wordnet.NOUN))\nprint(lemmatizer.lemmatize(\"troubling\", wordnet.VERB))\nprint(lemmatizer.lemmatize(\"troubled\", wordnet.VERB))","metadata":{"id":"CzTULvebDJgV","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"2426921def0946a4ab5ab7c5ae50623c","outputId":"e091a299-d51c-4af8-c5c4-5142c9ead3fd","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":3355,"user_tz":240,"timestamp":1653261265067},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"Lancaster Stemmer\ntroubl\ntroubl\ntroubl\nWordNet Lemmatizer\ntrouble\ntrouble\ntrouble\n"}],"execution_count":null},{"cell_type":"markdown","source":"Se puede ver que obtenemos una raíz de significado de Lemmatizer, mientras que Stemmer simplemente recorta y extrae la primera parte importante de la palabra.","metadata":{"id":"MfnEC2oUDQc5","cell_id":"f951894c6b294543aae1936652aaa5b4","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Creacion de modelo de ML","metadata":{"id":"EKYImfc-DSu4","cell_id":"65531152936e49aa83e271720692034b","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"amazon_reviews[['Text','Score','Sentiment_rating']].head(5)","metadata":{"id":"fdRlAQcWDWEf","colab":{"height":206,"base_uri":"https://localhost:8080/"},"cell_id":"b511e2cc367340dfbe99a96ca6225700","outputId":"a61604b0-1910-45c1-8e1b-1327aaaea8ce","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":363,"user_tz":240,"timestamp":1653261297738},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":" Text Score Sentiment_rating\n0 I have bought several of the Vitality canned d... 5 1\n1 Product arrived labeled as Jumbo Salted Peanut... 1 0\n2 This is a confection that has been around a fe... 4 1\n3 If you are looking for the secret ingredient i... 2 0\n4 Great taffy at a great price. There was a wid... 5 1","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
TextScoreSentiment_rating
0I have bought several of the Vitality canned d...51
1Product arrived labeled as Jumbo Salted Peanut...10
2This is a confection that has been around a fe...41
3If you are looking for the secret ingredient i...20
4Great taffy at a great price. There was a wid...51
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":29}],"execution_count":null},{"cell_type":"markdown","source":"Las variables independientes o características del modelo se derivan del texto de revisión. Previamente, discutimos cómo podemos usar **n-grams** para crear características, y específicamente cómo la bolsa de palabras es la interpretación más simple de estos n-gramas, sin tener en cuenta el orden y el contexto por completo y solo enfocándonos en la frecuencia/recuento. Usemos eso como punto de partida.","metadata":{"id":"I-0HWHdbDYwZ","cell_id":"70e9c97f8488438aaa09615b856db41c","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Bag of words\n\nCountVectorizer es una clase de Python que da cuenta automáticamente de ciertos pasos de preprocesamiento, como la eliminación de palabras vacías, la derivación, la creación de n-gramas y la tokenización de palabras:","metadata":{"id":"qENuMxTvDen5","cell_id":"aeec9334fc344c66a474c1cc3c40ec31","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"# Creacion de metodo para stemming\nfrom nltk.stem import PorterStemmer\n\nstemmer = PorterStemmer()\nanalyzer = CountVectorizer().build_analyzer()\n\ndef stemmed_words(doc):\n return (stemmer.stem(w) for w in analyzer(doc))","metadata":{"id":"esyrY_TqDdst","cell_id":"93c0e92a9df642abab69f9702787acd6","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Usemos esto para crear una bolsa de palabras de las reseñas, excluyendo las palabras irrelevantes que identificamos anteriormente:","metadata":{"id":"8OHClEiJDr9_","cell_id":"fa68a4b1adc2441c9b225ac0313ce0e8","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"# Creacion de un objeto tipo CountVectorizer\nbow_counts = CountVectorizer(\n tokenizer=word_tokenize,\n stop_words=noise_words,\n ngram_range=(1, 1)\n)","metadata":{"id":"7wV6wAd8DupW","cell_id":"826feace271946ae96b96a19c97244c5","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Una vez que se prepara la bolsa de palabras, el conjunto de datos debe dividirse en conjuntos de entrenamiento y de prueba. También podríamos dividir los datos después de vectorizarlos, pero es útil dividir los datos lo antes posible en el proceso. Esto significa que una vez que hemos generado nuestras predicciones, podemos compararlas más fácilmente con los textos originales, antes de que hayan sido preprocesadas y vectorizadas.","metadata":{"id":"0I--fC0SDz3o","cell_id":"dbdfd8b242844732b0e7c84136b64bff","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"reviews_train, reviews_test = train_test_split(amazon_reviews, test_size=0.2, random_state=0)","metadata":{"id":"3-D0n379D5_G","cell_id":"e25ad7f80e5c4ccba5e8332e4d79e733","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"X_train_bow = bow_counts.fit_transform(reviews_train.reviews_text_new)\nX_test_bow = bow_counts.transform(reviews_test.reviews_text_new)","metadata":{"id":"3EkmtCUeD7Tu","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"3aac2465c764490aa6914a25a73407e4","outputId":"6cc420e5-bf80-4b42-d98a-e6af702a4ad5","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":6641,"user_tz":240,"timestamp":1653261456737},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stderr","text":"/usr/local/lib/python3.7/dist-packages/sklearn/feature_extraction/text.py:401: UserWarning: Your stop_words may be inconsistent with your preprocessing. Tokenizing the stop words generated tokens [\"'d\", \"'ll\", \"'re\", \"'s\", \"'ve\", 'might', 'must', \"n't\", 'need', 'sha', 'wo'] not in stop_words.\n % sorted(inconsistent)\n"}],"execution_count":null},{"cell_type":"markdown","source":"Tengan en cuenta que llamamos a ```fit_transform``` para vectorizar nuestro conjunto de entrenamiento y ```transform``` para vectorizar nuestro conjunto de prueba. Esto genera las asignaciones de vectorización solo en los datos del conjunto de entrenamiento, que es una restricción a la que nos enfrentaríamos en un problema del mundo real (no tener acceso a los datos de prueba durante el tiempo de entrenamiento).\n\nPor lo tanto, puede haber algunas palabras en el conjunto de prueba que no sabemos cómo vectorizar y se omitirán.","metadata":{"id":"8q-tk94fD_Vu","cell_id":"fb6fe1363e264c71aafdf89e35a1bf18","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"y_train_bow = reviews_train['Sentiment_rating']\ny_test_bow = reviews_test['Sentiment_rating']","metadata":{"id":"mu-ClINID76n","cell_id":"18e17c065e1249278c87670f06827829","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"y_test_bow.value_counts() / y_test_bow.shape[0]","metadata":{"id":"KE-P_hwcEJzc","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"aafcf5ca68bb4fe7bcf8907b870ce65f","outputId":"88d11217-3aa5-488f-bfce-48456300cca0","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":383,"user_tz":240,"timestamp":1653261511710},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"1 0.847921\n0 0.152079\nName: Sentiment_rating, dtype: float64"},"metadata":{},"execution_count":35}],"execution_count":null},{"cell_type":"markdown","source":"Los datos de prueba contienen 84% de opiniones positivas. El modelo de predicción más simple que podríamos pensar sería uno que prediga \"positivo\" para cada entrada. Llamaríamos a esto un modelo \"ingenuo\", y constituye una línea de base útil. En este caso, dicho modelo obtendría un 84 % de precisión, por lo que podemos considerarlo como una puntuación de referencia que nuestro modelo de aprendizaje automático debe superar.","metadata":{"id":"i3RSbM62ENeq","cell_id":"1d4f7d35ee42490f932e69e8fbe6f215","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"## Modelo de regresion logistica","metadata":{"id":"b5GGWWW3EYyR","cell_id":"b83d927d224043fba9866384f7be8280","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"# Entrenar el modelo\nlr_model_all = LogisticRegression(C=1, solver=\"liblinear\")\nlr_model_all.fit(X_train_bow, y_train_bow)\n\n# Predecir el output\ntest_pred_lr_prob = lr_model_all.predict_proba(X_test_bow)\ntest_pred_lr_all = lr_model_all.predict(X_test_bow)\n\nprint(\"F1 score: \", f1_score(y_test_bow, test_pred_lr_all))\nprint(\"Accuracy: \", accuracy_score(y_test_bow, test_pred_lr_all) * 100)","metadata":{"id":"_H_xrYMjEcBs","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"bbb2e00c459e4fb7b2cec24ec42fbe79","outputId":"148a6c87-bee1-460a-9c15-b4a781a1bdf4","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":901,"user_tz":240,"timestamp":1653261600835},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"F1 score: 0.942550505050505\nAccuracy: 90.04376367614879\n"}],"execution_count":null},{"cell_type":"code","source":"test_pred_lr_prob","metadata":{"id":"B9g9Z2AFEijY","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"a0067f829fd049f39b7dd9dd74e563fb","outputId":"00e19690-216e-4b4a-fb8d-7057e46345aa","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":382,"user_tz":240,"timestamp":1653261619777},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"array([[7.97531898e-01, 2.02468102e-01],\n [2.70697065e-09, 9.99999997e-01],\n [5.15753584e-02, 9.48424642e-01],\n ...,\n [1.23713200e-04, 9.99876287e-01],\n [9.96585322e-01, 3.41467779e-03],\n [9.18040817e-02, 9.08195918e-01]])"},"metadata":{},"execution_count":38}],"execution_count":null},{"cell_type":"code","source":"probabilities = [each[1] for each in test_pred_lr_prob]","metadata":{"id":"c3pDuetkEmfq","cell_id":"370ac5caa3bb43faaa2287baefdcdbda","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"predictions = pd.DataFrame()\npredictions['Text'] = reviews_test['Text']\npredictions['Actual_Score'] = reviews_test['Score']\npredictions['Sentiment_rating'] = reviews_test['Sentiment_rating']\npredictions['Predicted_sentiment'] = test_pred_lr_all\npredictions['Predicted_probability'] = probabilities","metadata":{"id":"yhK1q_skEoqv","cell_id":"baf0e4099be94b109eb9217cbd0cac23","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"predictions.head(5)","metadata":{"id":"tk36qT2QEruF","colab":{"height":424,"base_uri":"https://localhost:8080/"},"cell_id":"5de9695d91454190bb7eb09c68ca3bb8","outputId":"327719ad-3e1f-48f7-f1a5-f47f3e37a27e","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":9,"user_tz":240,"timestamp":1653261649104},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":" Text Actual_Score \\\n7419 These truffles are not as great as they are ma... 2 \n4315 Ok--I just bought BIsquick GF and the first th... 4 \n1781 I love these chips so much... and I can't get ... 5 \n962 I purchased this because I read that it was a ... 2 \n7289 I do like the idea of these crackers for glute... 2 \n\n Sentiment_rating Predicted_sentiment Predicted_probability \n7419 0 0 0.202468 \n4315 1 1 1.000000 \n1781 1 1 0.948425 \n962 0 0 0.363864 \n7289 0 1 0.954062 ","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
TextActual_ScoreSentiment_ratingPredicted_sentimentPredicted_probability
7419These truffles are not as great as they are ma...2000.202468
4315Ok--I just bought BIsquick GF and the first th...4111.000000
1781I love these chips so much... and I can't get ...5110.948425
962I purchased this because I read that it was a ...2000.363864
7289I do like the idea of these crackers for glute...2010.954062
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":41}],"execution_count":null},{"cell_type":"code","source":"accuracy_score(predictions['Sentiment_rating'], predictions['Predicted_sentiment'])","metadata":{"id":"yOp3mDlXEuJA","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"50d2db3274a24431a9cfd41b74e44cd5","outputId":"628f8ea3-bc69-403e-9b1c-f3ca2bc99a74","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":440,"user_tz":240,"timestamp":1653261658529},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"0.9004376367614879"},"metadata":{},"execution_count":42}],"execution_count":null},{"cell_type":"markdown","source":"En la columna ```Predicted_probability```, puede ver la confianza que tenía el modelo en sus predicciones, siendo las probabilidades muy cercanas a 0 predicciones de sentimientos negativos muy confiables y las probabilidades muy cercanas a 1, predicciones de sentimientos positivos muy confiables.\n\nUtilice esta información para encontrar el caso en el que el modelo tenía más confianza al predecir que una reseña tenía un sentimiento negativo cuando la puntuación real era positiva.\n\nMire el texto y escriba algunas oraciones de análisis sobre por qué cree que el modelo se equivocó","metadata":{"id":"OBKZYoZtF4qw","cell_id":"1adae403f3a84e9e967b5d77cb7c5659","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"predictions[\n predictions['Predicted_sentiment'] != predictions['Sentiment_rating']\n].sort_values(by=[\"Predicted_probability\"]).head(3)","metadata":{"id":"VN0O6ei8GuOS","colab":{"height":292,"base_uri":"https://localhost:8080/"},"cell_id":"6bb93e4e466a4e3b8a32a8dd80d6a8ee","outputId":"634e4f98-9049-4b75-b963-ead16bd8e03b","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":520,"user_tz":240,"timestamp":1653262183193},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":" Text Actual_Score \\\n7692 My little girl (6 months) can't stomach any fo... 5 \n4166 The first time I tried this product to make pa... 4 \n6612 When I opened the package, I only found 11 pac... 4 \n\n Sentiment_rating Predicted_sentiment Predicted_probability \n7692 1 0 0.000719 \n4166 1 0 0.002114 \n6612 1 0 0.006040 ","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
TextActual_ScoreSentiment_ratingPredicted_sentimentPredicted_probability
7692My little girl (6 months) can't stomach any fo...5100.000719
4166The first time I tried this product to make pa...4100.002114
6612When I opened the package, I only found 11 pac...4100.006040
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":43}],"execution_count":null},{"cell_type":"code","source":"predictions.loc[7692].values","metadata":{"id":"S-G3sENDGu_i","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"a8f24770efdc4fc498e128e0bbb63ba9","outputId":"e759240a-10d6-4421-8a25-7530360a980e","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":417,"user_tz":240,"timestamp":1653262189852},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"array([\"My little girl (6 months) can't stomach any formula. Even Nutramigen, at $35 per can made her colicky and sick (accompanied with the worst smelling spit up). The expensive formulas smelled bad, and tasted worse, I hated giving them to her. Goats milk was recommended to me. Her doctor's only objection was that it was low in Folic Acid and tended to dehydrate babies. Well, Meyenberg Powdered Milk is fortified with Folic Acid, so dilute it with a little extra water and you're in business! I now have the happiest infant you've ever seen. Enfamil can take that nasty chemical soup they call formula and shove it, I'll never go back.

You can get the formula Recipe at:
[...]

There have been concerns about arsenic in rice syrup lately, so I have been using Lyle's Golden Syrup, and she likes it fine (Also it's the best pancake syrup you'll ever have). I am going to try barley malt syrup next, it is chemically more similar to rice syrup.\",\n 5, 1, 0, 0.0007189017742100002], dtype=object)"},"metadata":{},"execution_count":44}],"execution_count":null},{"cell_type":"markdown","source":"Podemos ver que la reseña tiene un tono muy negativo y utiliza palabras que la modelo probablemente ha aprendido a asociar fuertemente con malas reseñas como \"caro\", \"malo\", \"peor\", \"odiado\", \"desagradable\", \" nunca\". Sin embargo, todos estos en realidad están dirigidos a un producto diferente. El autor dice que después de comprar este producto tiene el \"bebé más feliz\", pero eso no es lo suficientemente fuerte como para contrarrestar todos los aspectos negativos de la reseña.","metadata":{"id":"3zbOLxXOGxk_","cell_id":"e7f2af358123437d843a99b528d6cca6","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Modifique el conjunto de características en el modelo para incluir bigramas, trigramas y 4-grams. No elimine las palabras irrelevantes definidas anteriormente antes de presentarlas. (Sugerencia: establece ngram_range=(1,4).)\n\nAl mismo tiempo, experimente con el ajuste de hiperparámetros. Cambie el valor C del clasificador de regresión logística a 0,9.","metadata":{"id":"uq3urjaaHMV0","cell_id":"7bc3f6b91f0042d6b9a4e29d622eab78","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"# Cambios con respecto al código anterior\n# 1. Aumentar los n-gramas de solo tener 1 gramo a (1 gramo, 2 gramos, 3 gramos y 4 gramos)\n# 2. Incluir las palabras vacías en las características de la bolsa de palabras\nbow_counts = CountVectorizer(\n tokenizer=word_tokenize,\n ngram_range=(1,4)\n)\n\nX_train_bow = bow_counts.fit_transform(reviews_train.reviews_text_new)\nX_test_bow = bow_counts.transform(reviews_test.reviews_text_new)","metadata":{"id":"kEqVP5IuHSKk","cell_id":"79aafb0a428f4593a1760be7b744df3b","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# Observe el aumento de funciones con la inclusión de palabras vacías\nX_train_bow","metadata":{"id":"8IPlWx1OHYKg","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"d378fedcba5c4792b3ffd5183af4e07a","outputId":"aed3e843-9871-4f9c-d07f-8712cbdbe6bd","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":504,"user_tz":240,"timestamp":1653262371710},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":"<7310x1054259 sparse matrix of type ''\n\twith 2036674 stored elements in Compressed Sparse Row format>"},"metadata":{},"execution_count":46}],"execution_count":null},{"cell_type":"code","source":"# Cambios en la regresión logística\n# Cambio de la sanción de regularización por defecto de l2 a l1\n# Cambiando el parámetro de costo C a 0.9\nlr_model_all_new = LogisticRegression(C=0.9, solver=\"liblinear\")","metadata":{"id":"GatU304PHdRw","cell_id":"809c87dc229d4406bfa0dacc93e9976c","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# Entrenar el modelo\nlr_model_all_new.fit(X_train_bow, y_train_bow)\n\n# Predecir resultados\ntest_pred_lr_prob = lr_model_all_new.predict_proba(X_test_bow)\ntest_pred_lr_all = lr_model_all_new.predict(X_test_bow)\n\nprint(\"F1 score: \", f1_score(y_test_bow, test_pred_lr_all))\nprint(\"Accuracy: \", accuracy_score(y_test_bow, test_pred_lr_all) * 100)","metadata":{"id":"t-9RjVU3HjeM","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"b65406b875044b60b1b58ef77e5fd035","outputId":"9a7ed7cb-94ec-42eb-b871-2fdc46affd00","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":21013,"user_tz":240,"timestamp":1653262433168},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"F1 score: 0.9554579673776662\nAccuracy: 92.23194748358861\n"}],"execution_count":null},{"cell_type":"markdown","source":"La precisión ha saltado del 90% al 92,2%. Este es un ejemplo de lo que el simple ajuste de hiperparámetros y la modificación de características de entrada pueden hacer en el rendimiento general. Incluso podemos obtener características interpretables de esto en términos de lo que más contribuyó al sentimiento positivo y negativo.","metadata":{"id":"AU11TtV5Hp3J","cell_id":"8dc1d33fe9ef416d99e771498565dae6","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"lr_weights = pd.DataFrame(list(\n zip(\n bow_counts.get_feature_names(),\n lr_model_all_new.coef_[0])\n ),\n columns=['words','weights']\n)\n\nlr_weights.sort_values(['weights'],ascending = False)[:15]","metadata":{"id":"1zboEFrIHvC5","colab":{"height":575,"base_uri":"https://localhost:8080/"},"cell_id":"191acded7ca64ef2b7bfaffec6a46cfa","outputId":"ee875f03-e61a-4c28-e22a-7efba540a77f","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":6151,"user_tz":240,"timestamp":1653262454049},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stderr","text":"/usr/local/lib/python3.7/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function get_feature_names is deprecated; get_feature_names is deprecated in 1.0 and will be removed in 1.2. Please use get_feature_names_out instead.\n warnings.warn(msg, category=FutureWarning)\n"},{"output_type":"execute_result","data":{"text/plain":" words weights\n374800 great 1.382526\n256072 delicious 0.979694\n366512 good 0.864971\n677656 perfect 0.844286\n299760 excellent 0.838121\n855190 the best 0.826370\n534066 love 0.819486\n143456 best 0.813387\n593253 nice 0.780972\n536985 loves 0.660368\n1034499 wonderful 0.653160\n777859 smooth 0.642421\n309549 favorite 0.615238\n605055 not too 0.587947\n315915 find 0.585533","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
wordsweights
374800great1.382526
256072delicious0.979694
366512good0.864971
677656perfect0.844286
299760excellent0.838121
855190the best0.826370
534066love0.819486
143456best0.813387
593253nice0.780972
536985loves0.660368
1034499wonderful0.653160
777859smooth0.642421
309549favorite0.615238
605055not too0.587947
315915find0.585533
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":49}],"execution_count":null},{"cell_type":"code","source":"lr_weights.sort_values(['weights'],ascending = False)[-15:]","metadata":{"id":"jFRk5csSHxyj","colab":{"height":520,"base_uri":"https://localhost:8080/"},"cell_id":"889421d6317c427e8c56323ea00f52cf","outputId":"c4d2ae4c-01ea-4e1c-fd77-ffec093df08e","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":1023,"user_tz":240,"timestamp":1653262460288},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":" words weights\n1037701 worst -0.526030\n1007881 were -0.529210\n982174 very disappointed -0.540921\n120594 away -0.546831\n730653 return -0.547383\n801742 stick -0.561657\n121444 awful -0.570031\n997363 waste -0.591713\n265649 disappointing -0.598279\n820835 t -0.654783\n124042 bad -0.663200\n1003928 weak -0.702415\n294693 even -0.782674\n599002 not -0.975730\n265330 disappointed -1.121489","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
wordsweights
1037701worst-0.526030
1007881were-0.529210
982174very disappointed-0.540921
120594away-0.546831
730653return-0.547383
801742stick-0.561657
121444awful-0.570031
997363waste-0.591713
265649disappointing-0.598279
820835t-0.654783
124042bad-0.663200
1003928weak-0.702415
294693even-0.782674
599002not-0.975730
265330disappointed-1.121489
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":50}],"execution_count":null},{"cell_type":"markdown","source":"## Random forest","metadata":{"id":"u4WxBMCnHz-H","cell_id":"2fe00730b19c43798799e8d24ef4bc0c","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"rf_model_all = RandomForestClassifier(n_estimators=100)\n\n# Entrenamiento\nrf_model_all.fit(X_train_bow, y_train_bow)\n\n# predicciones\ntest_pred_lr_prob = rf_model_all.predict_proba(X_test_bow)\ntest_pred_lr_all = rf_model_all.predict(X_test_bow)","metadata":{"id":"8qAvpMnoH2zm","cell_id":"32075822458b42c1ac117742cdcad288","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"print(\"F1 score: \", f1_score(y_test_bow,test_pred_lr_all))\nprint(\"Accuracy: \", accuracy_score(y_test_bow,test_pred_lr_all)* 100)","metadata":{"id":"r2Pu1uwZH7Ew","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"2f2a3b27fcd24f5780b0131b7f39a0ee","outputId":"e267f797-86f6-4064-9df8-3c30e6906801","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":7,"user_tz":240,"timestamp":1653262557913},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"F1 score: 0.9281437125748503\nAccuracy: 86.87089715536105\n"}],"execution_count":null},{"cell_type":"markdown","source":"Esto no es tan bueno como la regresión logística. Podemos obtener los n-gramas que fueron más importantes para las predicciones de la siguiente manera:","metadata":{"id":"jpMWiD1gH-d8","cell_id":"8ecfa8d499db40bdb8459315e75263ec","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"feature_importances = pd.DataFrame(\n rf_model_all.feature_importances_,\n index=bow_counts.get_feature_names(),\n columns=['importance']\n)","metadata":{"id":"l-pQMcXKIA3O","cell_id":"8a715665945c4c1092c9a01f6a70fa98","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"feature_importances.sort_values(['importance'], ascending=False)[:10]","metadata":{"id":"_PP7hG-3IBXy","cell_id":"bd0e84490b8e44d5a7b1d6fca2e4c894","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"# Modelo TF-IDF\n\nPor supuesto, la bolsa de palabras no es la única forma de caracterizar el texto. Otro método, que mencionamos brevemente antes, es el método TF-IDF. Esto evalúa qué tan importante es una palabra para un documento dentro de una gran colección de documentos (es decir, corpus). La importancia aumenta proporcionalmente en función del número de veces que aparece una palabra en el documento, pero se compensa con la frecuencia de la palabra en el corpus.\n\nLa TF-IDF es el producto de dos términos. El primero calcula la frecuencia de término normalizada (TF); es decir, el número de veces que aparece una palabra en un documento dividido por el número total de palabras en ese documento. El segundo término es la Frecuencia Inversa de Documentos (IDF), calculada como el logaritmo del número de documentos en el corpus dividido por el número de documentos ","metadata":{"id":"2jl8v2WlIDHs","cell_id":"7b57dccafc024220aad5c5f969ea79b9","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Menos formalmente, ¿qué significa esto?\n\n- Si una palabra aparece con mucha frecuencia en un documento específico, es probable que sea significativa.\n- Si una palabra aparece con mucha frecuencia en casi todos los documentos del corpus, es poco probable que sea significativa.\n- Por tanto, una palabra que aparece a menudo en un documento pero rara vez en el resto del corpus merece especial atención.\n\nTF-IDF no solo cuenta cada palabra, sino que aplica una ponderación para que las palabras comunes reciban menos atención y las palabras raras reciban más.\n\nVolvamos a presentar nuestro conjunto original de revisiones basado en TF-IDF y dividamos las funciones resultantes en conjuntos de entrenamiento y prueba:","metadata":{"id":"iaDByDseIdYq","cell_id":"965957ed0b3d42f09157676bf19f2199","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"# Cree un vectorizador: aún alimentamos nuestras stopwords, aunque\n# estos son menos relevantes ahora ya que TF-IDF los ponderaría menos\n# de todas formas.\ntfidf_counts = TfidfVectorizer(\n tokenizer=word_tokenize,\n stop_words=noise_words,\n ngram_range=(1,1)\n)\n\nX_train_tfidf = tfidf_counts.fit_transform(reviews_train.reviews_text_new)\nX_test_tfidf = tfidf_counts.transform(reviews_test.reviews_text_new)","metadata":{"id":"Qy27CGjZInLg","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"07bd2108855946cd9e01b6318f669f61","outputId":"87f0fcfb-49cc-4316-c2ab-2e33c2026f30","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":4591,"user_tz":240,"timestamp":1653262703595},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stderr","text":"/usr/local/lib/python3.7/dist-packages/sklearn/feature_extraction/text.py:401: UserWarning: Your stop_words may be inconsistent with your preprocessing. Tokenizing the stop words generated tokens [\"'d\", \"'ll\", \"'re\", \"'s\", \"'ve\", 'might', 'must', \"n't\", 'need', 'sha', 'wo'] not in stop_words.\n % sorted(inconsistent)\n"}],"execution_count":null},{"cell_type":"markdown","source":"## Regresion logistica","metadata":{"id":"2mcp31kGIuN4","cell_id":"9ec55bccb02b458eaabd29ebdca94c3f","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"# Crear el clasificador\nlr_model_tf_idf = LogisticRegression(solver=\"liblinear\")\n\n# Entrenar\nlr_model_tf_idf.fit(X_train_tfidf, y_train_bow)\n\n# Predecir\ntest_pred_lr_prob = lr_model_tf_idf.predict_proba(X_test_tfidf)\ntest_pred_lr_all = lr_model_tf_idf.predict(X_test_tfidf)\n\n## Evaluar el modelo\nprint(\"F1 score: \",f1_score(y_test_bow, test_pred_lr_all))\nprint(\"Accuracy: \", accuracy_score(y_test_bow, test_pred_lr_all) * 100)","metadata":{"id":"h2fKRigVIxBs","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"2a4dca61b6ba49ccaedf9d9ba025cac8","outputId":"3372785a-1984-48f5-caff-266ec16bbdb3","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":535,"user_tz":240,"timestamp":1653262736670},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"F1 score: 0.9343419062027231\nAccuracy: 88.12910284463895\n"}],"execution_count":null},{"cell_type":"markdown","source":"Aquí hemos logrado una precisión del 88 % con TF-IDF en comparación con el 90 % con 1-gram. Es difícil saber exactamente por qué este algoritmo de vectorización más sofisticado conduce a peores resultados, pero podría ser que penalizar palabras que son comunes en todo el corpus genere una desventaja para este conjunto de datos en particular. TF-IDF suele ser útil cuando los datos de prueba son muy diferentes de los datos de entrenamiento, lo que permite que se desprioricen las palabras que solo son comunes en el conjunto de entrenamiento.","metadata":{"id":"A4uMiNZmI4Db","cell_id":"70a48335b7b04e74b6cd6e740526a13c","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Trate de aumentar la precisión del modelo por\n\n- configurando ```ngram_range=(1,4)``` en el Vectorizador\n- no eliminar las palabras irrelevantes de antemano en el Vectorizador\n- configurando C=10 en el clasificador LogisticRegression\n- estableciendo penalización=\"l1\" en el clasificador LogisticRegression","metadata":{"id":"SCsby16EJBGA","cell_id":"996f723d4b914c1492de8fca2db65bda","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"tfidf_counts = TfidfVectorizer(\n tokenizer=word_tokenize,\n ngram_range=(1,4)\n)\n\nX_train_tfidf = tfidf_counts.fit_transform(reviews_train.reviews_text_new)\nX_test_tfidf = tfidf_counts.transform(reviews_test.reviews_text_new)","metadata":{"id":"5UlnsuRcJKwQ","cell_id":"055ee761769d495ebf82d904d4278a09","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# definiendo la clase del modelo\nlr_model_tf_idf_new = LogisticRegression(solver=\"liblinear\", penalty='l1', C=10)\n\n# Entrenar\nlr_model_tf_idf_new.fit(X_train_tfidf, y_train_bow)\n\n# Predecir\ntest_pred_lr_prob = lr_model_tf_idf_new.predict_proba(X_test_tfidf)\ntest_pred_lr_all = lr_model_tf_idf_new.predict(X_test_tfidf)\n\n# Evaluar el modelo\nprint(\"F1 score: \",f1_score(y_test_bow, test_pred_lr_all))\nprint(\"Accuracy: \", accuracy_score(y_test_bow, test_pred_lr_all)*100)","metadata":{"id":"Ur44xFjOJNqx","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"fb1d712e52ec4809b7cc28a2fca7bcc1","outputId":"5dea35f6-3d59-4119-c32e-3355831d7291","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":5419,"user_tz":240,"timestamp":1653262866726},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"F1 score: 0.9470967741935484\nAccuracy: 91.02844638949672\n"}],"execution_count":null},{"cell_type":"markdown","source":"Esta es una mejora con respecto a nuestro resultado anterior, pero hicimos cuatro cambios al mismo tiempo, por lo que no sabemos cuáles ayudaron y cuánto.\n\nProbar diferentes hiperparámetros para mejorar su modelo se denomina ajuste de hiperparámetros y es un campo enorme por sí solo. Puede imaginar cómo ejecutar este modelo 16 veces, una vez con cada posible configuración de hiperparámetros que hemos probado, sería un poco complicado de seguir, ¡y esto es solo con 4 hiperparámetros y dos valores para cada uno! Con 100 o 1000 de hiperparámetros y 100 o 1000 de valores para cada uno, las combinaciones totales crecen muy rápidamente.\n\nPara ayudar con esto, scikit-learn proporciona la llamada funcionalidad de \"GridSearch\", donde puede configurar una canalización y especificar los rangos de hiperparámetros que desea \"buscar\". Scikit-learn probará cada combinación y entrenará y evaluará el modelo para cada caso, diciéndole cuál funcionó mejor.\n\nTambién podemos encontrar nuestras características más importantes nuevamente, como se muestra a continuación:","metadata":{"id":"VXgj1CPgJYg7","cell_id":"b482c250be2f4e9fa83733faf6071fe7","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"lr_weights = pd.DataFrame(\n list(\n zip(tfidf_counts.get_feature_names(), lr_model_tf_idf_new.coef_[0])\n ),\n columns=['words','weights']\n)\n\nlr_weights.sort_values(['weights'],ascending = False)[:10]","metadata":{"id":"lOnUb9NiJiEd","colab":{"height":418,"base_uri":"https://localhost:8080/"},"cell_id":"9db2346394f84314925733f083e8b7fd","outputId":"13471d7e-e549-4648-9870-603cddad4e60","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":3568,"user_tz":240,"timestamp":1653262930606},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stderr","text":"/usr/local/lib/python3.7/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function get_feature_names is deprecated; get_feature_names is deprecated in 1.0 and will be removed in 1.2. Please use get_feature_names_out instead.\n warnings.warn(msg, category=FutureWarning)\n"},{"output_type":"execute_result","data":{"text/plain":" words weights\n374800 great 51.682173\n677656 perfect 46.173501\n855190 the best 44.691956\n256072 delicious 39.746053\n63261 amazing 38.903587\n131810 be disappointed 38.802913\n299760 excellent 37.935929\n777859 smooth 36.007994\n685128 pleased 33.385577\n593253 nice 32.073194","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
wordsweights
374800great51.682173
677656perfect46.173501
855190the best44.691956
256072delicious39.746053
63261amazing38.903587
131810be disappointed38.802913
299760excellent37.935929
777859smooth36.007994
685128pleased33.385577
593253nice32.073194
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":57}],"execution_count":null},{"cell_type":"code","source":"lr_weights.sort_values(['weights'],ascending = False)[-10:]","metadata":{"id":"GvmARJjeJkG2","colab":{"height":363,"base_uri":"https://localhost:8080/"},"cell_id":"5d5c525c3c4f44f7940555cb0bfe6cc3","outputId":"21600646-cc73-46d8-dc66-f2eeb7cbdc26","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":349,"user_tz":240,"timestamp":1653262933269},"deepnote_cell_type":"code"},"outputs":[{"output_type":"execute_result","data":{"text/plain":" words weights\n730653 return -34.726053\n599002 not -35.237744\n603641 not recommend -35.535403\n843696 than this -36.479892\n1037701 worst -37.469519\n601259 not for -38.382928\n1053388 yuck -41.851058\n605704 not worth -44.629182\n265330 disappointed -45.957117\n265649 disappointing -49.420275","text/html":"\n
\n
\n
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
wordsweights
730653return-34.726053
599002not-35.237744
603641not recommend-35.535403
843696than this-36.479892
1037701worst-37.469519
601259not for-38.382928
1053388yuck-41.851058
605704not worth-44.629182
265330disappointed-45.957117
265649disappointing-49.420275
\n
\n \n \n \n\n \n
\n
\n "},"metadata":{},"execution_count":58}],"execution_count":null},{"cell_type":"markdown","source":"# Modelo Word Embeddings","metadata":{"id":"Pu7WHTrpJl_8","cell_id":"acf848498593443cb88500a118e6f93a","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"El tipo final de caracterización que cubriremos son las incrustaciones de palabras (Word Embeddings). Este es un tipo de representación de palabras que permite que palabras con un significado similar tengan una representación similar. Al ser entrenados previamente en datos externos, como Wikipedia, las incrustaciones de palabras saben cuándo los conceptos están relacionados semánticamente; por ejemplo, los vectores para \"rey\" y \"reina\" se ubicarían uno cerca del otro, aunque no hay sintaxis ni similitud en ortografía entre estas palabras.\n\nEs este enfoque para representar palabras y documentos el que puede considerarse uno de los avances clave del aprendizaje profundo en los desafiantes problemas de procesamiento del lenguaje natural.\n\nHay muchos conjuntos de datos de incrustaciones de palabras previamente entrenadas que están disponibles gratuitamente, o puede entrenar las suyas propias. Algunos de los avances más importantes en esta área incluyen Word2Vec, que se convirtió en el modelo a seguir de la NPL, y otros enfoques integrados como Glove, ELMo y BERT.","metadata":{"id":"cvsz0rEbJrmb","cell_id":"9ad4244b64b745d5a7b2b8fdd6b9d2fe","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Existen diferentes métodos para aprender incrustaciones de palabras: Word2Vec, GloVe, FastText. **Word2Vec** utiliza una red neuronal superficial y es de dos tipos; _CBOW_ y _Skip Gram_. **GloVe** es un algoritmo de aprendizaje no supervisado para obtener representaciones vectoriales de palabras. El entrenamiento se realiza en estadísticas globales agregadas de coocurrencia palabra-palabra de un corpus, y las representaciones resultantes muestran subestructuras lineales interesantes del espacio vectorial de palabras. **[fastText](https://fasttext.cc/)** es una biblioteca para aprender incrustaciones de palabras y clasificación de texto creada por el laboratorio de investigación de IA de Facebook.","metadata":{"id":"h8N02FZJJ-VF","cell_id":"bab87570bf5d4bb38a717f2161183421","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"**Porque usar Word Embeddings en lugar de Bag of words o TF-IDF**\nCada palabra está representada por un vector de valor real, que generalmente tiene decenas o cientos de dimensiones. Esto contrasta con los miles o millones de dimensiones requeridas para las representaciones de palabras dispersas. Por lo tanto, las incrustaciones de palabras pueden reducir drásticamente la cantidad de dimensiones requeridas para representar un documento de texto:","metadata":{"id":"2gzgZC5UKMyS","cell_id":"41d5ccae70744d7da57f46ea4339946b","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"import gensim\n# Cargar una incrustación de palabras de guante preentrenada que se entrena en un conjunto de datos de Twitter\n# Esta palabra incrustada tiene 200 dimensiones, lo que significa que cada palabra está representada\n# por un vector de 200 dimensiones.\nmodel = gensim.models.KeyedVectors.load_word2vec_format(\n os.path.join(os.getcwd(), 'glove.twitter.27B.200d_out.txt'),\n binary=False,\n unicode_errors='ignore'\n)","metadata":{"id":"fWaGepmSKVwl","cell_id":"76d8db05cd344d9f82b8c8e4cc6b4286","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Teníamos aproximadamente 18000 tokens distintos para características de 1-grams en la representación de la bolsa de palabras, pero solo tendrán 200 dimensiones en esta inserción de palabras. ¡Esta es una gran diferencia!\n\nAdemás, las incrustaciones de palabras capturan el contexto y la semántica de las oraciones, ya que cada representación de vector de palabra se basa en su significado contextual.\n\nA continuación se muestra la representación vectorial de \"comida\" y \"genial\":","metadata":{"id":"OlmzuwOFKjRM","cell_id":"3e024d61feaf400d903c80c468c95ec9","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"print(\"El embedding para food es\", len(model['food']), \"dimensional\")\n\nmodel['food']","metadata":{"id":"OCU3QFnmKfXr","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"f85c346491bd4be2a5df763ace217e94","outputId":"602a23a0-bcde-44a3-e0b5-d7646b6da310","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":9,"user_tz":240,"timestamp":1653263230563},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"El embedding para food es 200 dimensional\n"},{"output_type":"execute_result","data":{"text/plain":"array([-6.9175e-01, -1.4259e-01, 3.8653e-01, -2.3141e-01, -2.0408e-01,\n -2.1565e-01, 7.7839e-01, 2.2689e-03, -7.2446e-02, -6.0134e-01,\n -4.2400e-01, -5.7140e-01, -8.4249e-01, 1.5947e-01, -1.2899e-01,\n 5.9032e-01, -1.3632e-01, -6.6478e-01, -1.9557e-01, -8.2453e-01,\n -1.3177e-01, 1.3514e-01, -7.3214e-01, 4.8200e-01, 4.3505e-01,\n 1.6676e+00, -1.8275e-01, -1.0007e-01, 3.7003e-01, 1.0411e-01,\n -8.8115e-01, -9.7733e-04, -2.9459e-01, -7.3869e-02, -4.0103e-01,\n -4.6626e-01, 2.3253e-01, 2.7776e-01, 4.0754e-01, -4.5051e-02,\n -1.9468e-01, -2.9230e-01, -3.4642e-01, -4.9286e-01, 1.0467e-01,\n 7.2143e-01, 5.9596e-01, 5.3495e-01, 3.8788e-02, -1.4406e-01,\n -5.2248e-02, -6.8292e-01, -1.0080e-01, -1.2961e-01, -2.6006e-02,\n 1.4836e-01, 3.2417e-02, 1.3997e-01, 8.3943e-03, -2.3139e-01,\n -1.8000e-01, -3.1689e-01, 2.3606e-01, 1.8237e-01, 4.3933e-01,\n -3.2313e-01, -2.1512e-03, -4.4172e-01, 4.1011e-01, 1.7174e-01,\n -8.6405e-01, -3.9674e-01, 4.4175e-01, 5.9300e-01, 1.8982e-01,\n -2.9646e-02, -3.4041e-01, -3.3708e-02, 7.3449e-01, 4.5300e-01,\n -2.7855e-02, -1.8993e-02, 3.8107e-01, -5.6606e-02, 1.4864e-02,\n 3.1518e-01, -3.2304e-01, -2.7439e-01, 6.1900e-02, 3.2886e-01,\n 1.5138e-01, 5.3268e-01, -1.6616e-01, -2.3076e-01, -9.6515e-02,\n 4.5991e-01, -5.1475e-01, 1.0297e-01, -4.0225e-02, 5.6679e-01,\n 3.1027e-01, 1.5679e-01, -2.5897e-01, 4.6312e-01, 2.2561e-01,\n -3.9300e-01, -3.9593e-01, 4.4001e-01, 3.7176e-01, 1.4747e-02,\n -1.9193e-01, -2.2478e-01, -1.2665e-01, -3.4982e-01, 5.0847e-01,\n 3.1720e-01, 1.2942e-01, -6.2695e-01, 5.8675e-01, 4.1040e-02,\n 1.8835e-01, -2.2626e-01, -1.1744e-01, 5.1429e-03, 7.2058e-02,\n -4.9525e-01, 4.4159e-01, 8.6225e-01, 7.6765e-02, -9.7908e-02,\n 6.8383e-02, 3.0596e-01, 3.7980e-01, 1.1563e-01, -6.1020e-01,\n -6.8107e-01, 3.2723e-02, 2.5346e-01, 3.5334e-01, 2.5407e-01,\n -4.6516e-01, 4.8858e-01, 3.9032e-01, -8.1296e-01, -6.9780e-01,\n -1.2542e-01, 7.9234e-02, 1.2918e-01, -1.1048e-01, 8.9312e-03,\n 3.6999e-01, 3.0116e-01, -4.6578e+00, -4.4493e-03, 2.0313e-02,\n -5.0215e-02, -2.0646e-01, -3.7321e-02, -5.1779e-02, 6.6986e-02,\n -5.8853e-01, 7.1753e-01, 4.2784e-02, 1.6667e-03, -2.6193e-01,\n 5.8214e-01, -1.0513e+00, -3.0341e-02, 7.3892e-01, -1.8003e-01,\n -1.1104e-01, 3.0846e-01, 4.4027e-01, -8.4080e-02, -2.6251e-01,\n -3.8733e-01, -2.6630e-01, 1.9655e-01, 5.3812e-02, -2.4456e-01,\n -7.8868e-01, -7.1843e-01, 7.0593e-02, -1.9051e-01, 2.5553e-01,\n -1.3786e-01, 1.2942e-01, 4.5864e-01, 5.5462e-01, 8.2104e-01,\n -2.5049e-01, -3.3623e-01, 1.8491e-01, -4.8235e-01, 3.1425e-01,\n 2.4499e-01, -2.4404e-01, 8.0309e-02, 3.4060e-01, 7.0451e-01],\n dtype=float32)"},"metadata":{},"execution_count":61}],"execution_count":null},{"cell_type":"code","source":"print(\"El embedding para great es\", len(model['great']), \"dimensional\")\n\nmodel['great']","metadata":{"id":"1g4FvfwRKtp5","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"5dd2a45527e54a7e96e917adad731eeb","outputId":"5a9927cf-325c-46d1-b6cc-824796f5e00a","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":410,"user_tz":240,"timestamp":1653263239565},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"El embedding para great es 200 dimensional\n"},{"output_type":"execute_result","data":{"text/plain":"array([ 1.0751e-01, 1.5958e-01, 1.3332e-01, 1.6642e-01, -3.2737e-02,\n 1.7592e-01, 7.2395e-01, 1.1713e-01, -3.5036e-01, -4.2937e-01,\n -4.0925e-01, -2.5761e-01, -1.0264e+00, -1.0014e-01, 5.5390e-02,\n 2.0413e-01, 1.2807e-01, -2.6337e-02, -6.9719e-02, -3.6193e-02,\n -1.9917e-01, 3.9437e-02, -9.2358e-02, 2.6981e-01, -2.0951e-01,\n 1.5455e+00, -2.8123e-01, 3.2046e-01, 4.5545e-01, -3.8841e-02,\n -1.7369e-01, -2.3251e-01, -5.9551e-02, 2.3250e-01, 4.4214e-01,\n 3.3666e-01, 3.9352e-02, -1.2462e-01, -2.9317e-01, -4.8857e-02,\n 6.9021e-01, 7.1279e-02, 1.0252e-01, 1.6122e-01, -2.3536e-01,\n 6.2724e-02, 2.0222e-01, 5.0234e-02, -1.1611e-01, 2.8909e-02,\n -1.1109e-01, -5.0241e-02, -5.9063e-01, -8.8747e-02, 5.1444e-01,\n -1.3715e-01, 1.7194e-01, -8.3657e-02, 9.6333e-02, -9.7063e-02,\n 3.4003e-03, -7.0180e-02, -5.9588e-01, -2.8264e-01, 1.2529e-01,\n 2.4359e-01, -4.9082e-01, -4.2533e-02, 2.2158e-01, -2.1491e-01,\n -4.2101e-02, 2.3359e-01, 3.1978e-01, 3.5063e-01, 6.1748e-01,\n -1.0197e-01, 5.3357e-01, -3.6005e-01, -1.7212e-02, 1.6645e-01,\n 8.9432e-01, 2.7322e-02, 3.0683e-01, 1.9715e-02, 6.0516e-01,\n 4.1085e-01, 5.5945e-01, -8.4501e-02, 3.5933e-01, 1.0216e-01,\n 2.6675e-01, -6.0445e-01, -1.0513e-01, -1.9248e-01, 2.9150e-01,\n -1.0537e-01, 5.2671e-01, 2.3763e-01, -1.3640e-01, -6.1029e-02,\n 1.0081e-01, 7.4541e-02, -1.4899e-01, -2.2301e-01, -1.3653e-02,\n 4.0192e-02, 5.5821e-03, -2.9936e-02, 2.7338e-02, 5.9412e-01,\n -1.0302e-01, 9.0319e-02, 3.1055e-01, 6.3336e-01, 2.9762e-01,\n -8.4671e-02, -1.2552e-01, -6.3930e-01, 3.8613e-01, 6.6371e-01,\n 5.1345e-01, 2.0719e-01, 2.1100e-01, 1.4579e-01, -7.3321e-02,\n -7.0593e-01, -6.2578e-02, -2.5470e-01, 1.1986e-01, 1.6102e-01,\n 3.2958e-02, -2.4159e-01, -2.5708e-01, 3.2051e-01, -1.1569e-01,\n 6.7540e-03, -1.1688e-01, -3.6158e-02, -6.5320e-01, 4.9560e-01,\n -3.9429e-02, -1.8395e-01, 2.3295e-01, 5.4128e-01, 2.4568e-02,\n -1.9862e-01, 2.1041e-01, 9.3798e-02, 8.3096e-03, -6.1551e-02,\n 2.3262e-01, -4.2756e-02, -5.3511e+00, 3.0604e-01, 3.3578e-01,\n -3.6771e-01, 5.6225e-01, -8.2341e-02, 2.9809e-01, 2.5189e-01,\n -4.6203e-01, 1.0452e-01, -3.9540e-01, 3.6961e-01, 1.3093e-01,\n 1.6653e-01, -3.1915e-01, 1.6974e-01, 4.2575e-01, 3.6420e-01,\n 3.7175e-01, -1.9450e-01, 6.2702e-02, 4.9775e-01, 3.1842e-02,\n -6.4072e-02, 7.6183e-02, -5.9534e-01, 3.1731e-01, -2.8254e-01,\n 1.5987e-01, -9.2750e-02, -4.1426e-02, 7.5799e-02, 9.5740e-03,\n -2.1532e-01, -3.1419e-01, -1.5144e-01, -4.6584e-01, -1.1069e-01,\n -4.0130e-01, 3.9266e-02, 8.1880e-01, -4.2955e-02, 2.1698e-01,\n -6.0347e-02, 3.3431e-01, -9.9549e-02, -1.8156e-01, -8.5143e-02],\n dtype=float32)"},"metadata":{},"execution_count":62}],"execution_count":null},{"cell_type":"markdown","source":"Como discutimos, el poder de las incrustaciones de palabras es que las palabras que tienen un significado similar están más juntas en el espacio vectorial. Podemos demostrar esto mirando la distancia del coseno entre algunos pares de palabras de la siguiente manera:","metadata":{"id":"z6zkIX7UKzlV","cell_id":"3e7784e97dc840c99e49e217129c90ce","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"def print_similarity(word1, word2, model):\n v1 = model[word1]\n v2 = model[word2]\n similarity = np.dot(v1, v2) / (np.linalg.norm(v1) * np.linalg.norm(v2))\n print(f\"{word1} y {word2} son {round(similarity * 100)}% similar\")\n\nprint_similarity(\"cat\", \"dog\", model)\nprint_similarity(\"good\", \"bad\", model)\nprint_similarity(\"great\", \"good\", model)\nprint_similarity(\"grass\", \"model\", model)","metadata":{"id":"1LG9xirxK2ET","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"85f03dbcf2d94897bbf486971eca1957","outputId":"51d1839e-be29-4fd0-a8a8-4645dd785a53","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":415,"user_tz":240,"timestamp":1653263273770},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"cat y dog son 83% similar\ngood y bad son 80% similar\ngreat y good son 87% similar\ngrass y model son 11% similar\n"}],"execution_count":null},{"cell_type":"markdown","source":"Aquí, \"significado similar\" se define vagamente como \"usado en contextos similares\". Debido a que hay muchos ejemplos como \"Acaricié a mi gato\" y \"Acaricié a mi perro\" donde estas palabras se usan en contextos similares, se consideran muy similares. También hay muchas oraciones en las que \"bueno\" y \"malo\" pueden intercambiarse y la oración puede seguir siendo válida, por lo que, aunque las consideremos \"opuestas\", nuestro modelo las considerará similares. \"hierba\" y \"modelo\" no tienen casi nada que ver entre sí, por lo que están muy separados.\n\nPara encontrar el vector de una reseña completa, obtenemos el vector de cada palabra de la reseña por separado y tomamos un promedio simple","metadata":{"id":"NyPpFZbQK8Ra","cell_id":"6614aa2d6acd4875bbc5228ac6da2d57","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"Calcule el vector para cada revisión individual en el conjunto de datos.","metadata":{"id":"Axse7IKFLC0v","cell_id":"818f13c4644b44698a7e7107526ef537","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"review_embeddings = []\n\nfor each_review in amazon_reviews.reviews_text_new:\n review_average = np.zeros(model.vector_size)\n count_val = 0\n \n for each_word in word_tokenize(each_review):\n\n # Cambiar a \"if True\" para remover stopwords del promedio de embeddings\n if False:\n if(each_word.lower() in noise_words):\n print(each_word.lower())\n continue\n \n if(each_word.lower() in model):\n review_average += model[each_word.lower()]\n count_val += 1\n \n review_embeddings.append(list(review_average/count_val))","metadata":{"id":"RQbosshgLHXN","cell_id":"37617c382ef24e1da7822691a75a3cf8","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Cargar sus propios vectores de palabras es excelente para comprender cómo funcionan, pero también existen bibliotecas de nivel superior que abstraen código como el que se muestra arriba. En la industria, una biblioteca de NLP ampliamente utilizada es [SpaCy] (https://spacy.io/). Esta biblioteca le permite extraer de manera eficiente las incrustaciones de palabras de los textos y realizar operaciones de alto nivel sobre ellos.\n\nConvirtamos la lista de representaciones vectoriales para cada revisión en un DataFrame y dividámoslo en conjuntos de entrenamiento y prueba:","metadata":{"id":"xeD-CD9zLRqG","cell_id":"c165449f4f40430cb2ea3cb33f2b8889","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"embedding_data = pd.DataFrame(review_embeddings)\nembedding_data = embedding_data.fillna(0)","metadata":{"id":"YFztCDMZLWbz","cell_id":"9c699427347b498fac92aed7636e9361","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"X_train_embed, X_test_embed, y_train_embed, y_test_embed = train_test_split(\n embedding_data,\n amazon_reviews.Sentiment_rating,\n test_size=0.2,\n random_state=0\n)","metadata":{"id":"PKDP6ZYzLX_3","cell_id":"bb0836a6283e4373bac8b5f3cc29af46","deepnote_cell_type":"code"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Ahora apliquemos el modelo de regresion logistica a nuestras word embeddings ","metadata":{"id":"SPWhQqYKLZX2","cell_id":"f1eb8bbb243340159a9ea81fae0f810f","deepnote_cell_type":"markdown"}},{"cell_type":"code","source":"lr_model = LogisticRegression(penalty=\"l1\", C=10, solver=\"liblinear\")\nlr_model.fit(X_train_embed, y_train_embed)\ntest_pred_lr_prob = lr_model.predict_proba(X_test_embed)\ntest_pred_lr_all = lr_model.predict(X_test_embed)\n\nprint(\"F1 score: \", f1_score(y_test_embed, test_pred_lr_all))\nprint(\"Accuracy: \", accuracy_score(y_test_embed, test_pred_lr_all)*100)","metadata":{"id":"ecE_YyKuLc2-","colab":{"base_uri":"https://localhost:8080/"},"cell_id":"0b862de8975e4c99a7510c3c98ba9a96","outputId":"edd8451e-0640-4247-d307-692fe0410513","executionInfo":{"user":{"userId":"09471607480253994520","displayName":"David Francisco Bustos Usta"},"status":"ok","elapsed":14370,"user_tz":240,"timestamp":1653263441295},"deepnote_cell_type":"code"},"outputs":[{"output_type":"stream","name":"stdout","text":"F1 score: 0.9381703470031545\nAccuracy: 89.27789934354486\n"}],"execution_count":null},{"cell_type":"markdown","source":"Desafortunadamente, esto no es tan bueno como las representaciones de bolsa de palabras o TF-IDF. Además, aunque las incrustaciones de palabras fueron realmente efectivas para reducir el número total de dimensiones, adolece del problema de la interpretabilidad. Esto significa que es muy difícil para nosotros incluso diagnosticar qué está causando su bajo rendimiento.\n\nSin embargo, ¿recuerda cómo \"bueno\" y \"malo\" estaban juntos en el espacio vectorial? Esta es una de las razones por las que las incrustaciones de palabras pueden no funcionar tan bien para el análisis de sentimientos en conjuntos de datos más pequeños: las incrustaciones de palabras son buenas para usar \"conocimiento\" del mundo externo (latente en las incrustaciones preentrenadas) para inferir información adicional sobre un conjunto de datos más pequeño, pero en el caso del análisis de sentimientos, esto podría hacer más daño que bien al combinar palabras \"similares\" que en realidad están muy separadas para una tarea de análisis de sentimientos.\n\n**En nuestro caso, la creación de características usando TF-IDF nos dio una precisión del 92 % con características muy interpretables. Esta es una buena combinación, por lo que consideramos que este es el mejor modelo para nosotros aquí**.\n\nTenga en cuenta que para un experimento real, habríamos dividido nuestro conjunto de datos en tres partes, no solo en dos. Cuando se ejecuta un experimento varias veces con diferentes parámetros, es casi seguro que algunos resultados serán mejores simplemente por casualidad, y es mala ciencia seleccionar el modelo con mejor desempeño después de docenas o cientos de ejecuciones.\n\nPara evitar este problema, los datos deben dividirse en conjuntos de \"entrenamiento\", \"prueba\" y \"validación\". El conjunto de \"prueba\" debe reservarse al comienzo del experimento y nunca mirarse. El modelo debe ajustarse utilizando el conjunto de \"validación\".\n\nSolo una vez que el experimentador esté satisfecho con el modelo al mejorar el rendimiento en el conjunto de validación, se debe ejecutar el modelo en el conjunto de prueba y esos resultados finales se toman como los resultados finales del experimento.","metadata":{"id":"aLgmytl8LjYx","cell_id":"54762887d23c436d901c6c1a517709bb","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Conclusiones","metadata":{"id":"mACgQsCSLzu9","cell_id":"23d9dc3b75a04e39a8047c400b20845b","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"En este caso, limpiamos y destacamos un conjunto de datos de reseñas de Amazon y construimos algunos modelos de clasificación en estas características para predecir el sentimiento. Vimos que la bolsa de palabras y TF-IDF brindaban características interpretables, mientras que las incrustaciones de palabras realmente no. Al aumentar el conjunto de n-gramas que usamos de 1-gram a 4-grams, pudimos obtener la precisión de nuestro modelo de regresión logística hasta en un 92 %.","metadata":{"id":"hPvVo2MuL04X","cell_id":"e14dfd80116844828ad355d448250ce0","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"# Para llevar","metadata":{"id":"8IUVh266MA_L","cell_id":"524495d87068404da0df83776a14d1cd","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"La construcción de modelos de aprendizaje automático en texto es una disciplina muy complicada. Algunas cosas importantes a tener en cuenta son las siguientes:\n\n1. Aunque existen diferentes tipos de preprocesamiento involucrados en los datos textuales, no todo tiene que aplicarse en cada caso. Por ejemplo, cuando se trata de mensajes de texto, los caracteres especiales pueden representar información importante y no es necesario eliminarlos. Además, las mayúsculas pueden significar que alguien está enojado y representan gritos, por lo que las mayúsculas y las minúsculas pueden representar información valiosa. En otras situaciones, es más valioso normalizarlas.\n\n2. El ajuste de hiperparámetros en los modelos de aprendizaje automático es un paso muy importante y, si bien los hiperparámetros predeterminados funcionan bien en muchos casos, a menudo se puede obtener un rendimiento adicional al ajustarlos. Se deben probar diferentes conjuntos de parámetros para ver qué contribuye al mejor modelo.\n\n3. Cada tarea de clasificación en NLP es diferente, pero el proceso a seguir es similar al que hicimos en este caso: **discutir los datos -> crear características a partir del texto -> entrenar modelos -> evaluar modelos**.","metadata":{"id":"iN7hN-ZNMCUX","cell_id":"bbe64c90cf2f4da397bc96668991a84d","deepnote_cell_type":"markdown"}},{"cell_type":"markdown","source":"\nCreated in deepnote.com \nCreated in Deepnote","metadata":{"created_in_deepnote_cell":true,"deepnote_cell_type":"markdown"}}],"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"Clase 15 - Introduccion a NLP Clase 2.ipynb","provenance":[],"authorship_tag":"ABX9TyMJOIDex6+Fa4Bm/eZnYXYV","collapsed_sections":["eB-FMmvvOh90","RrNKtx8EPrr1","Lq1Cev75Qjy5","27Owz6clRAe2","WPyUzf66RL-V","qzhKZ2n_TEv4","asrthaXMT1j0"]},"deepnote":{},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"},"deepnote_notebook_id":"c75de597e02c486db219573ea95d9faa","deepnote_execution_queue":[]}}