{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "**Important: This notebook will only work with fastai-0.7.x. Do not try to run any fastai-1.x code from this path in the repository because it will load fastai-0.7.x**" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Intro to Random Forests" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## About this course" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Teaching approach" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This course is being taught by Jeremy Howard, and was developed by Jeremy along with Rachel Thomas. Rachel has been dealing with a life-threatening illness so will not be teaching as originally planned this year.\n", "\n", "Jeremy has worked in a number of different areas - feel free to ask about anything that he might be able to help you with at any time, even if not directly related to the current topic:\n", "\n", "- Management consultant (McKinsey; AT Kearney)\n", "- Self-funded startup entrepreneur (Fastmail: first consumer synchronized email; Optimal Decisions: first optimized insurance pricing)\n", "- VC-funded startup entrepreneur: (Kaggle; Enlitic: first deep-learning medical company)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "I'll be using a *top-down* teaching method, which is different from how most math courses operate. Typically, in a *bottom-up* approach, you first learn all the separate components you will be using, and then you gradually build them up into more complex structures. The problems with this are that students often lose motivation, don't have a sense of the \"big picture\", and don't know what they'll need.\n", "\n", "If you took the fast.ai deep learning course, that is what we used. You can hear more about my teaching philosophy [in this blog post](http://www.fast.ai/2016/10/08/teaching-philosophy/) or [in this talk](https://vimeo.com/214233053).\n", "\n", "Harvard Professor David Perkins has a book, [Making Learning Whole](https://www.amazon.com/Making-Learning-Whole-Principles-Transform/dp/0470633719) in which he uses baseball as an analogy. We don't require kids to memorize all the rules of baseball and understand all the technical details before we let them play the game. Rather, they start playing with a just general sense of it, and then gradually learn more rules/details as time goes on.\n", "\n", "All that to say, don't worry if you don't understand everything at first! You're not supposed to. We will start using some \"black boxes\" such as random forests that haven't yet been explained in detail, and then we'll dig into the lower level details later.\n", "\n", "To start, focus on what things DO, not what they ARE." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Your practice" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "People learn by:\n", "1. **doing** (coding and building)\n", "2. **explaining** what they've learned (by writing or helping others)\n", "\n", "Therefore, we suggest that you practice these skills on Kaggle by:\n", "1. Entering competitions (*doing*)\n", "2. Creating Kaggle kernels (*explaining*)\n", "\n", "It's OK if you don't get good competition ranks or any kernel votes at first - that's totally normal! Just try to keep improving every day, and you'll see the results over time." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "To get better at technical writing, study the top ranked Kaggle kernels from past competitions, and read posts from well-regarded technical bloggers. Some good role models include:\n", "\n", "- [Peter Norvig](http://nbviewer.jupyter.org/url/norvig.com/ipython/ProbabilityParadox.ipynb) (more [here](http://norvig.com/ipython/))\n", "- [Stephen Merity](https://smerity.com/articles/2017/deepcoder_and_ai_hype.html)\n", "- [Julia Evans](https://codewords.recurse.com/issues/five/why-do-neural-networks-think-a-panda-is-a-vulture) (more [here](https://jvns.ca/blog/2014/08/12/what-happens-if-you-write-a-tcp-stack-in-python/))\n", "- [Julia Ferraioli](http://blog.juliaferraioli.com/2016/02/exploring-world-using-vision-twilio.html)\n", "- [Edwin Chen](http://blog.echen.me/2014/10/07/moving-beyond-ctr-better-recommendations-through-human-evaluation/)\n", "- [Slav Ivanov](https://blog.slavv.com/picking-an-optimizer-for-style-transfer-86e7b8cba84b) (fast.ai student)\n", "- [Brad Kenstler](https://hackernoon.com/non-artistic-style-transfer-or-how-to-draw-kanye-using-captain-picards-face-c4a50256b814) (fast.ai and USF MSAN student)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Books" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The more familiarity you have with numeric programming in Python, the better. If you're looking to improve in this area, we strongly suggest Wes McKinney's [Python for Data Analysis, 2nd ed](https://www.amazon.com/Python-Data-Analysis-Wrangling-IPython/dp/1491957662/ref=asap_bc?ie=UTF8).\n", "\n", "For machine learning with Python, we recommend:\n", "\n", "- [Introduction to Machine Learning with Python](https://www.amazon.com/Introduction-Machine-Learning-Andreas-Mueller/dp/1449369413): From one of the scikit-learn authors, which is the main library we'll be using\n", "- [Python Machine Learning: Machine Learning and Deep Learning with Python, scikit-learn, and TensorFlow, 2nd Edition](https://www.amazon.com/Python-Machine-Learning-scikit-learn-TensorFlow/dp/1787125939/ref=dp_ob_title_bk): New version of a very successful book. A lot of the new material however covers deep learning in Tensorflow, which isn't relevant to this course\n", "- [Hands-On Machine Learning with Scikit-Learn and TensorFlow](https://www.amazon.com/Hands-Machine-Learning-Scikit-Learn-TensorFlow/dp/1491962291/ref=pd_lpo_sbs_14_t_0?_encoding=UTF8&psc=1&refRID=MBV2QMFH3EZ6B3YBY40K)\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Syllabus in brief" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Depending on time and class interests, we'll cover something like (not necessarily in this order):\n", "\n", "- Train vs test\n", " - Effective validation set construction\n", "- Trees and ensembles\n", " - Creating random forests\n", " - Interpreting random forests\n", "- What is ML? Why do we use it?\n", " - What makes a good ML project?\n", " - Structured vs unstructured data\n", " - Examples of failures/mistakes\n", "- Feature engineering\n", " - Domain specific - dates, URLs, text\n", " - Embeddings / latent factors\n", "- Regularized models trained with SGD\n", " - GLMs, Elasticnet, etc (NB: see what James covered)\n", "- Basic neural nets\n", " - PyTorch\n", " - Broadcasting, Matrix Multiplication\n", " - Training loop, backpropagation\n", "- KNN\n", "- CV / bootstrap (Diabetes data set?)\n", "- Ethical considerations" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Skip:\n", "\n", "- Dimensionality reduction\n", "- Interactions\n", "- Monitoring training\n", "- Collaborative filtering\n", "- Momentum and LR annealing\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Imports" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "%load_ext autoreload\n", "%autoreload 2\n", "\n", "%matplotlib inline" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/home/jhoward/anaconda3/lib/python3.6/site-packages/sklearn/ensemble/weight_boosting.py:29: DeprecationWarning: numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n", " from numpy.core.umath_tests import inner1d\n" ] } ], "source": [ "from fastai.imports import *\n", "from fastai.structured import *\n", "\n", "from pandas_summary import DataFrameSummary\n", "from sklearn.ensemble import RandomForestRegressor, RandomForestClassifier\n", "from IPython.display import display\n", "\n", "from sklearn import metrics" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "PATH = \"data/bulldozers/\"" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Data%20Dictionary.xlsx\t\t Test.csv\t Train.csv\r\n", "Machine_Appendix.csv\t\t tmp\t\t Valid.csv\r\n", "median_benchmark.csv\t\t TrainAndValid.7z ValidSolution.csv\r\n", "models\t\t\t\t TrainAndValid.csv\r\n", "random_forest_benchmark_test.csv TrainAndValid.zip\r\n" ] } ], "source": [ "!ls {PATH}" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Introduction to *Blue Book for Bulldozers*" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## About..." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### ...our teaching" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "At fast.ai we have a distinctive [teaching philosophy](http://www.fast.ai/2016/10/08/teaching-philosophy/) of [\"the whole game\"](https://www.amazon.com/Making-Learning-Whole-Principles-Transform/dp/0470633719/ref=sr_1_1?ie=UTF8&qid=1505094653). This is different from how most traditional math & technical courses are taught, where you have to learn all the individual elements before you can combine them (Harvard professor David Perkins call this *elementitis*), but it is similar to how topics like *driving* and *baseball* are taught. That is, you can start driving without [knowing how an internal combustion engine works](https://medium.com/towards-data-science/thoughts-after-taking-the-deeplearning-ai-courses-8568f132153), and children begin playing baseball before they learn all the formal rules." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### ...our approach to machine learning" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Most machine learning courses will throw at you dozens of different algorithms, with a brief technical description of the math behind them, and maybe a toy example. You're left confused by the enormous range of techniques shown and have little practical understanding of how to apply them.\n", "\n", "The good news is that modern machine learning can be distilled down to a couple of key techniques that are of very wide applicability. Recent studies have shown that the vast majority of datasets can be best modeled with just two methods:\n", "\n", "- *Ensembles of decision trees* (i.e. Random Forests and Gradient Boosting Machines), mainly for structured data (such as you might find in a database table at most companies)\n", "- *Multi-layered neural networks learnt with SGD* (i.e. shallow and/or deep learning), mainly for unstructured data (such as audio, vision, and natural language)\n", "\n", "In this course we'll be doing a deep dive into random forests, and simple models learnt with SGD. You'll be learning about gradient boosting and deep learning in part 2." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### ...this dataset" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We will be looking at the Blue Book for Bulldozers Kaggle Competition: \"The goal of the contest is to predict the sale price of a particular piece of heavy equiment at auction based on it's usage, equipment type, and configuration. The data is sourced from auction result postings and includes information on usage and equipment configurations.\"\n", "\n", "This is a very common type of dataset and prediciton problem, and similar to what you may see in your project or workplace." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### ...Kaggle Competitions" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Kaggle is an awesome resource for aspiring data scientists or anyone looking to improve their machine learning skills. There is nothing like being able to get hands-on practice and receiving real-time feedback to help you improve your skills.\n", "\n", "Kaggle provides:\n", "\n", "1. Interesting data sets\n", "2. Feedback on how you're doing\n", "3. A leader board to see what's good, what's possible, and what's state-of-art.\n", "4. Blog posts by winning contestants share useful tips and techniques." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## The data" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Look at the data" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Kaggle provides info about some of the fields of our dataset; on the [Kaggle Data info](https://www.kaggle.com/c/bluebook-for-bulldozers/data) page they say the following:\n", "\n", "For this competition, you are predicting the sale price of bulldozers sold at auctions. The data for this competition is split into three parts:\n", "\n", "- **Train.csv** is the training set, which contains data through the end of 2011.\n", "- **Valid.csv** is the validation set, which contains data from January 1, 2012 - April 30, 2012. You make predictions on this set throughout the majority of the competition. Your score on this set is used to create the public leaderboard.\n", "- **Test.csv** is the test set, which won't be released until the last week of the competition. It contains data from May 1, 2012 - November 2012. Your score on the test set determines your final rank for the competition.\n", "\n", "The key fields are in train.csv are:\n", "\n", "- SalesID: the unique identifier of the sale\n", "- MachineID: the unique identifier of a machine. A machine can be sold multiple times\n", "- saleprice: what the machine sold for at auction (only provided in train.csv)\n", "- saledate: the date of the sale" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "*Question*\n", "\n", "What stands out to you from the above description? What needs to be true of our training and validation sets?" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df_raw = pd.read_csv(f'{PATH}Train.csv', low_memory=False, \n", " parse_dates=[\"saledate\"])" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In any sort of data science work, it's **important to look at your data**, to make sure you understand the format, how it's stored, what type of values it holds, etc. Even if you've read descriptions about your data, the actual data may not be what you expect." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "def display_all(df):\n", " with pd.option_context(\"display.max_rows\", 1000, \"display.max_columns\", 1000): \n", " display(df)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "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", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
401120401121401122401123401124
SalesID63333366333337633333863333416333342
SalePrice10500110001150090007750
MachineID18407021830472188765919035701926965
ModelID2143921439214392143521435
datasource149149149149149
auctioneerID11122
YearMade20052005200520052005
MachineHoursCurrentMeterNaNNaNNaNNaNNaN
UsageBandNaNNaNNaNNaNNaN
saledate2011-11-02 00:00:002011-11-02 00:00:002011-11-02 00:00:002011-10-25 00:00:002011-10-25 00:00:00
fiModelDesc35NX235NX235NX230NX30NX
fiBaseModel3535353030
fiSecondaryDescNXNXNXNXNX
fiModelSeries222NaNNaN
fiModelDescriptorNaNNaNNaNNaNNaN
ProductSizeMiniMiniMiniMiniMini
fiProductClassDescHydraulic Excavator, Track - 3.0 to 4.0 Metric...Hydraulic Excavator, Track - 3.0 to 4.0 Metric...Hydraulic Excavator, Track - 3.0 to 4.0 Metric...Hydraulic Excavator, Track - 2.0 to 3.0 Metric...Hydraulic Excavator, Track - 2.0 to 3.0 Metric...
stateMarylandMarylandMarylandFloridaFlorida
ProductGroupTEXTEXTEXTEXTEX
ProductGroupDescTrack ExcavatorsTrack ExcavatorsTrack ExcavatorsTrack ExcavatorsTrack Excavators
Drive_SystemNaNNaNNaNNaNNaN
EnclosureEROPSEROPSEROPSEROPSEROPS
ForksNaNNaNNaNNaNNaN
Pad_TypeNaNNaNNaNNaNNaN
Ride_ControlNaNNaNNaNNaNNaN
StickNaNNaNNaNNaNNaN
TransmissionNaNNaNNaNNaNNaN
TurbochargedNaNNaNNaNNaNNaN
Blade_ExtensionNaNNaNNaNNaNNaN
Blade_WidthNaNNaNNaNNaNNaN
Enclosure_TypeNaNNaNNaNNaNNaN
Engine_HorsepowerNaNNaNNaNNaNNaN
HydraulicsAuxiliaryStandardAuxiliaryStandardStandard
PushblockNaNNaNNaNNaNNaN
RipperNaNNaNNaNNaNNaN
ScarifierNaNNaNNaNNaNNaN
Tip_ControlNaNNaNNaNNaNNaN
Tire_SizeNaNNaNNaNNaNNaN
CouplerNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or Unspecified
Coupler_SystemNaNNaNNaNNaNNaN
Grouser_TracksNaNNaNNaNNaNNaN
Hydraulics_FlowNaNNaNNaNNaNNaN
Track_TypeSteelSteelSteelSteelSteel
Undercarriage_Pad_WidthNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or Unspecified
Stick_LengthNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or Unspecified
ThumbNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or Unspecified
Pattern_ChangerNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or UnspecifiedNone or Unspecified
Grouser_TypeDoubleDoubleDoubleDoubleDouble
Backhoe_MountingNaNNaNNaNNaNNaN
Blade_TypeNaNNaNNaNNaNNaN
Travel_ControlsNaNNaNNaNNaNNaN
Differential_TypeNaNNaNNaNNaNNaN
Steering_ControlsNaNNaNNaNNaNNaN
\n", "
" ], "text/plain": [ " 401120 \\\n", "SalesID 6333336 \n", "SalePrice 10500 \n", "MachineID 1840702 \n", "ModelID 21439 \n", "datasource 149 \n", "auctioneerID 1 \n", "YearMade 2005 \n", "MachineHoursCurrentMeter NaN \n", "UsageBand NaN \n", "saledate 2011-11-02 00:00:00 \n", "fiModelDesc 35NX2 \n", "fiBaseModel 35 \n", "fiSecondaryDesc NX \n", "fiModelSeries 2 \n", "fiModelDescriptor NaN \n", "ProductSize Mini \n", "fiProductClassDesc Hydraulic Excavator, Track - 3.0 to 4.0 Metric... \n", "state Maryland \n", "ProductGroup TEX \n", "ProductGroupDesc Track Excavators \n", "Drive_System NaN \n", "Enclosure EROPS \n", "Forks NaN \n", "Pad_Type NaN \n", "Ride_Control NaN \n", "Stick NaN \n", "Transmission NaN \n", "Turbocharged NaN \n", "Blade_Extension NaN \n", "Blade_Width NaN \n", "Enclosure_Type NaN \n", "Engine_Horsepower NaN \n", "Hydraulics Auxiliary \n", "Pushblock NaN \n", "Ripper NaN \n", "Scarifier NaN \n", "Tip_Control NaN \n", "Tire_Size NaN \n", "Coupler None or Unspecified \n", "Coupler_System NaN \n", "Grouser_Tracks NaN \n", "Hydraulics_Flow NaN \n", "Track_Type Steel \n", "Undercarriage_Pad_Width None or Unspecified \n", "Stick_Length None or Unspecified \n", "Thumb None or Unspecified \n", "Pattern_Changer None or Unspecified \n", "Grouser_Type Double \n", "Backhoe_Mounting NaN \n", "Blade_Type NaN \n", "Travel_Controls NaN \n", "Differential_Type NaN \n", "Steering_Controls NaN \n", "\n", " 401121 \\\n", "SalesID 6333337 \n", "SalePrice 11000 \n", "MachineID 1830472 \n", "ModelID 21439 \n", "datasource 149 \n", "auctioneerID 1 \n", "YearMade 2005 \n", "MachineHoursCurrentMeter NaN \n", "UsageBand NaN \n", "saledate 2011-11-02 00:00:00 \n", "fiModelDesc 35NX2 \n", "fiBaseModel 35 \n", "fiSecondaryDesc NX \n", "fiModelSeries 2 \n", "fiModelDescriptor NaN \n", "ProductSize Mini \n", "fiProductClassDesc Hydraulic Excavator, Track - 3.0 to 4.0 Metric... \n", "state Maryland \n", "ProductGroup TEX \n", "ProductGroupDesc Track Excavators \n", "Drive_System NaN \n", "Enclosure EROPS \n", "Forks NaN \n", "Pad_Type NaN \n", "Ride_Control NaN \n", "Stick NaN \n", "Transmission NaN \n", "Turbocharged NaN \n", "Blade_Extension NaN \n", "Blade_Width NaN \n", "Enclosure_Type NaN \n", "Engine_Horsepower NaN \n", "Hydraulics Standard \n", "Pushblock NaN \n", "Ripper NaN \n", "Scarifier NaN \n", "Tip_Control NaN \n", "Tire_Size NaN \n", "Coupler None or Unspecified \n", "Coupler_System NaN \n", "Grouser_Tracks NaN \n", "Hydraulics_Flow NaN \n", "Track_Type Steel \n", "Undercarriage_Pad_Width None or Unspecified \n", "Stick_Length None or Unspecified \n", "Thumb None or Unspecified \n", "Pattern_Changer None or Unspecified \n", "Grouser_Type Double \n", "Backhoe_Mounting NaN \n", "Blade_Type NaN \n", "Travel_Controls NaN \n", "Differential_Type NaN \n", "Steering_Controls NaN \n", "\n", " 401122 \\\n", "SalesID 6333338 \n", "SalePrice 11500 \n", "MachineID 1887659 \n", "ModelID 21439 \n", "datasource 149 \n", "auctioneerID 1 \n", "YearMade 2005 \n", "MachineHoursCurrentMeter NaN \n", "UsageBand NaN \n", "saledate 2011-11-02 00:00:00 \n", "fiModelDesc 35NX2 \n", "fiBaseModel 35 \n", "fiSecondaryDesc NX \n", "fiModelSeries 2 \n", "fiModelDescriptor NaN \n", "ProductSize Mini \n", "fiProductClassDesc Hydraulic Excavator, Track - 3.0 to 4.0 Metric... \n", "state Maryland \n", "ProductGroup TEX \n", "ProductGroupDesc Track Excavators \n", "Drive_System NaN \n", "Enclosure EROPS \n", "Forks NaN \n", "Pad_Type NaN \n", "Ride_Control NaN \n", "Stick NaN \n", "Transmission NaN \n", "Turbocharged NaN \n", "Blade_Extension NaN \n", "Blade_Width NaN \n", "Enclosure_Type NaN \n", "Engine_Horsepower NaN \n", "Hydraulics Auxiliary \n", "Pushblock NaN \n", "Ripper NaN \n", "Scarifier NaN \n", "Tip_Control NaN \n", "Tire_Size NaN \n", "Coupler None or Unspecified \n", "Coupler_System NaN \n", "Grouser_Tracks NaN \n", "Hydraulics_Flow NaN \n", "Track_Type Steel \n", "Undercarriage_Pad_Width None or Unspecified \n", "Stick_Length None or Unspecified \n", "Thumb None or Unspecified \n", "Pattern_Changer None or Unspecified \n", "Grouser_Type Double \n", "Backhoe_Mounting NaN \n", "Blade_Type NaN \n", "Travel_Controls NaN \n", "Differential_Type NaN \n", "Steering_Controls NaN \n", "\n", " 401123 \\\n", "SalesID 6333341 \n", "SalePrice 9000 \n", "MachineID 1903570 \n", "ModelID 21435 \n", "datasource 149 \n", "auctioneerID 2 \n", "YearMade 2005 \n", "MachineHoursCurrentMeter NaN \n", "UsageBand NaN \n", "saledate 2011-10-25 00:00:00 \n", "fiModelDesc 30NX \n", "fiBaseModel 30 \n", "fiSecondaryDesc NX \n", "fiModelSeries NaN \n", "fiModelDescriptor NaN \n", "ProductSize Mini \n", "fiProductClassDesc Hydraulic Excavator, Track - 2.0 to 3.0 Metric... \n", "state Florida \n", "ProductGroup TEX \n", "ProductGroupDesc Track Excavators \n", "Drive_System NaN \n", "Enclosure EROPS \n", "Forks NaN \n", "Pad_Type NaN \n", "Ride_Control NaN \n", "Stick NaN \n", "Transmission NaN \n", "Turbocharged NaN \n", "Blade_Extension NaN \n", "Blade_Width NaN \n", "Enclosure_Type NaN \n", "Engine_Horsepower NaN \n", "Hydraulics Standard \n", "Pushblock NaN \n", "Ripper NaN \n", "Scarifier NaN \n", "Tip_Control NaN \n", "Tire_Size NaN \n", "Coupler None or Unspecified \n", "Coupler_System NaN \n", "Grouser_Tracks NaN \n", "Hydraulics_Flow NaN \n", "Track_Type Steel \n", "Undercarriage_Pad_Width None or Unspecified \n", "Stick_Length None or Unspecified \n", "Thumb None or Unspecified \n", "Pattern_Changer None or Unspecified \n", "Grouser_Type Double \n", "Backhoe_Mounting NaN \n", "Blade_Type NaN \n", "Travel_Controls NaN \n", "Differential_Type NaN \n", "Steering_Controls NaN \n", "\n", " 401124 \n", "SalesID 6333342 \n", "SalePrice 7750 \n", "MachineID 1926965 \n", "ModelID 21435 \n", "datasource 149 \n", "auctioneerID 2 \n", "YearMade 2005 \n", "MachineHoursCurrentMeter NaN \n", "UsageBand NaN \n", "saledate 2011-10-25 00:00:00 \n", "fiModelDesc 30NX \n", "fiBaseModel 30 \n", "fiSecondaryDesc NX \n", "fiModelSeries NaN \n", "fiModelDescriptor NaN \n", "ProductSize Mini \n", "fiProductClassDesc Hydraulic Excavator, Track - 2.0 to 3.0 Metric... \n", "state Florida \n", "ProductGroup TEX \n", "ProductGroupDesc Track Excavators \n", "Drive_System NaN \n", "Enclosure EROPS \n", "Forks NaN \n", "Pad_Type NaN \n", "Ride_Control NaN \n", "Stick NaN \n", "Transmission NaN \n", "Turbocharged NaN \n", "Blade_Extension NaN \n", "Blade_Width NaN \n", "Enclosure_Type NaN \n", "Engine_Horsepower NaN \n", "Hydraulics Standard \n", "Pushblock NaN \n", "Ripper NaN \n", "Scarifier NaN \n", "Tip_Control NaN \n", "Tire_Size NaN \n", "Coupler None or Unspecified \n", "Coupler_System NaN \n", "Grouser_Tracks NaN \n", "Hydraulics_Flow NaN \n", "Track_Type Steel \n", "Undercarriage_Pad_Width None or Unspecified \n", "Stick_Length None or Unspecified \n", "Thumb None or Unspecified \n", "Pattern_Changer None or Unspecified \n", "Grouser_Type Double \n", "Backhoe_Mounting NaN \n", "Blade_Type NaN \n", "Travel_Controls NaN \n", "Differential_Type NaN \n", "Steering_Controls NaN " ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "display_all(df_raw.tail().T)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "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", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
countuniquetopfreqfirstlastmeanstdmin25%50%75%max
SalesID401125NaNNaNNaNNaNNaN1.91971e+069090211.13925e+061.41837e+061.63942e+062.24271e+066.33334e+06
SalePrice401125NaNNaNNaNNaNNaN31099.723036.94750145002400040000142000
MachineID401125NaNNaNNaNNaNNaN1.2179e+0644099201.0887e+061.27949e+061.46807e+062.48633e+06
ModelID401125NaNNaNNaNNaNNaN6889.76221.782832594604872437198
datasource401125NaNNaNNaNNaNNaN134.6668.96224121132132136172
auctioneerID380989NaNNaNNaNNaNNaN6.5560416.9768012499
YearMade401125NaNNaNNaNNaNNaN1899.16291.79710001985199520002013
MachineHoursCurrentMeter142765NaNNaNNaNNaNNaN3457.9627590.300030252.4833e+06
UsageBand696393Medium33985NaNNaNNaNNaNNaNNaNNaNNaNNaN
saledate40112539192009-02-16 00:00:0019321989-01-17 00:00:002011-12-30 00:00:00NaNNaNNaNNaNNaNNaNNaN
fiModelDesc4011254999310G5039NaNNaNNaNNaNNaNNaNNaNNaNNaN
fiBaseModel401125195058019798NaNNaNNaNNaNNaNNaNNaNNaNNaN
fiSecondaryDesc263934175C43235NaNNaNNaNNaNNaNNaNNaNNaNNaN
fiModelSeries56908122II13202NaNNaNNaNNaNNaNNaNNaNNaNNaN
fiModelDescriptor71919139L15875NaNNaNNaNNaNNaNNaNNaNNaNNaN
ProductSize1903506Medium62274NaNNaNNaNNaNNaNNaNNaNNaNNaN
fiProductClassDesc40112574Backhoe Loader - 14.0 to 15.0 Ft Standard Digg...56166NaNNaNNaNNaNNaNNaNNaNNaNNaN
state40112553Florida63944NaNNaNNaNNaNNaNNaNNaNNaNNaN
ProductGroup4011256TEX101167NaNNaNNaNNaNNaNNaNNaNNaNNaN
ProductGroupDesc4011256Track Excavators101167NaNNaNNaNNaNNaNNaNNaNNaNNaN
Drive_System1043614Two Wheel Drive46139NaNNaNNaNNaNNaNNaNNaNNaNNaN
Enclosure4008006OROPS173932NaNNaNNaNNaNNaNNaNNaNNaNNaN
Forks1920772None or Unspecified178300NaNNaNNaNNaNNaNNaNNaNNaNNaN
Pad_Type791344None or Unspecified70614NaNNaNNaNNaNNaNNaNNaNNaNNaN
Ride_Control1486063No77685NaNNaNNaNNaNNaNNaNNaNNaNNaN
Stick791342Standard48829NaNNaNNaNNaNNaNNaNNaNNaNNaN
Transmission1832308Standard140328NaNNaNNaNNaNNaNNaNNaNNaNNaN
Turbocharged791342None or Unspecified75211NaNNaNNaNNaNNaNNaNNaNNaNNaN
Blade_Extension252192None or Unspecified24692NaNNaNNaNNaNNaNNaNNaNNaNNaN
Blade_Width25219614'9615NaNNaNNaNNaNNaNNaNNaNNaNNaN
Enclosure_Type252193None or Unspecified21923NaNNaNNaNNaNNaNNaNNaNNaNNaN
Engine_Horsepower252192No23937NaNNaNNaNNaNNaNNaNNaNNaNNaN
Hydraulics320570122 Valve141404NaNNaNNaNNaNNaNNaNNaNNaNNaN
Pushblock252192None or Unspecified19463NaNNaNNaNNaNNaNNaNNaNNaNNaN
Ripper1041374None or Unspecified83452NaNNaNNaNNaNNaNNaNNaNNaNNaN
Scarifier252302None or Unspecified12719NaNNaNNaNNaNNaNNaNNaNNaNNaN
Tip_Control252193None or Unspecified16207NaNNaNNaNNaNNaNNaNNaNNaNNaN
Tire_Size9471817None or Unspecified46339NaNNaNNaNNaNNaNNaNNaNNaNNaN
Coupler2139523None or Unspecified184582NaNNaNNaNNaNNaNNaNNaNNaNNaN
Coupler_System434582None or Unspecified40430NaNNaNNaNNaNNaNNaNNaNNaNNaN
Grouser_Tracks433622None or Unspecified40515NaNNaNNaNNaNNaNNaNNaNNaNNaN
Hydraulics_Flow433623Standard42784NaNNaNNaNNaNNaNNaNNaNNaNNaN
Track_Type991532Steel84880NaNNaNNaNNaNNaNNaNNaNNaNNaN
Undercarriage_Pad_Width9987219None or Unspecified79651NaNNaNNaNNaNNaNNaNNaNNaNNaN
Stick_Length9921829None or Unspecified78820NaNNaNNaNNaNNaNNaNNaNNaNNaN
Thumb992883None or Unspecified83093NaNNaNNaNNaNNaNNaNNaNNaNNaN
Pattern_Changer992183None or Unspecified90255NaNNaNNaNNaNNaNNaNNaNNaNNaN
Grouser_Type991533Double84653NaNNaNNaNNaNNaNNaNNaNNaNNaN
Backhoe_Mounting786722None or Unspecified78652NaNNaNNaNNaNNaNNaNNaNNaNNaN
Blade_Type7983310PAT38612NaNNaNNaNNaNNaNNaNNaNNaNNaN
Travel_Controls798347None or Unspecified69923NaNNaNNaNNaNNaNNaNNaNNaNNaN
Differential_Type694114Standard68073NaNNaNNaNNaNNaNNaNNaNNaNNaN
Steering_Controls693695Conventional68679NaNNaNNaNNaNNaNNaNNaNNaNNaN
\n", "
" ], "text/plain": [ " count unique \\\n", "SalesID 401125 NaN \n", "SalePrice 401125 NaN \n", "MachineID 401125 NaN \n", "ModelID 401125 NaN \n", "datasource 401125 NaN \n", "auctioneerID 380989 NaN \n", "YearMade 401125 NaN \n", "MachineHoursCurrentMeter 142765 NaN \n", "UsageBand 69639 3 \n", "saledate 401125 3919 \n", "fiModelDesc 401125 4999 \n", "fiBaseModel 401125 1950 \n", "fiSecondaryDesc 263934 175 \n", "fiModelSeries 56908 122 \n", "fiModelDescriptor 71919 139 \n", "ProductSize 190350 6 \n", "fiProductClassDesc 401125 74 \n", "state 401125 53 \n", "ProductGroup 401125 6 \n", "ProductGroupDesc 401125 6 \n", "Drive_System 104361 4 \n", "Enclosure 400800 6 \n", "Forks 192077 2 \n", "Pad_Type 79134 4 \n", "Ride_Control 148606 3 \n", "Stick 79134 2 \n", "Transmission 183230 8 \n", "Turbocharged 79134 2 \n", "Blade_Extension 25219 2 \n", "Blade_Width 25219 6 \n", "Enclosure_Type 25219 3 \n", "Engine_Horsepower 25219 2 \n", "Hydraulics 320570 12 \n", "Pushblock 25219 2 \n", "Ripper 104137 4 \n", "Scarifier 25230 2 \n", "Tip_Control 25219 3 \n", "Tire_Size 94718 17 \n", "Coupler 213952 3 \n", "Coupler_System 43458 2 \n", "Grouser_Tracks 43362 2 \n", "Hydraulics_Flow 43362 3 \n", "Track_Type 99153 2 \n", "Undercarriage_Pad_Width 99872 19 \n", "Stick_Length 99218 29 \n", "Thumb 99288 3 \n", "Pattern_Changer 99218 3 \n", "Grouser_Type 99153 3 \n", "Backhoe_Mounting 78672 2 \n", "Blade_Type 79833 10 \n", "Travel_Controls 79834 7 \n", "Differential_Type 69411 4 \n", "Steering_Controls 69369 5 \n", "\n", " top \\\n", "SalesID NaN \n", "SalePrice NaN \n", "MachineID NaN \n", "ModelID NaN \n", "datasource NaN \n", "auctioneerID NaN \n", "YearMade NaN \n", "MachineHoursCurrentMeter NaN \n", "UsageBand Medium \n", "saledate 2009-02-16 00:00:00 \n", "fiModelDesc 310G \n", "fiBaseModel 580 \n", "fiSecondaryDesc C \n", "fiModelSeries II \n", "fiModelDescriptor L \n", "ProductSize Medium \n", "fiProductClassDesc Backhoe Loader - 14.0 to 15.0 Ft Standard Digg... \n", "state Florida \n", "ProductGroup TEX \n", "ProductGroupDesc Track Excavators \n", "Drive_System Two Wheel Drive \n", "Enclosure OROPS \n", "Forks None or Unspecified \n", "Pad_Type None or Unspecified \n", "Ride_Control No \n", "Stick Standard \n", "Transmission Standard \n", "Turbocharged None or Unspecified \n", "Blade_Extension None or Unspecified \n", "Blade_Width 14' \n", "Enclosure_Type None or Unspecified \n", "Engine_Horsepower No \n", "Hydraulics 2 Valve \n", "Pushblock None or Unspecified \n", "Ripper None or Unspecified \n", "Scarifier None or Unspecified \n", "Tip_Control None or Unspecified \n", "Tire_Size None or Unspecified \n", "Coupler None or Unspecified \n", "Coupler_System None or Unspecified \n", "Grouser_Tracks None or Unspecified \n", "Hydraulics_Flow Standard \n", "Track_Type Steel \n", "Undercarriage_Pad_Width None or Unspecified \n", "Stick_Length None or Unspecified \n", "Thumb None or Unspecified \n", "Pattern_Changer None or Unspecified \n", "Grouser_Type Double \n", "Backhoe_Mounting None or Unspecified \n", "Blade_Type PAT \n", "Travel_Controls None or Unspecified \n", "Differential_Type Standard \n", "Steering_Controls Conventional \n", "\n", " freq first last \\\n", "SalesID NaN NaN NaN \n", "SalePrice NaN NaN NaN \n", "MachineID NaN NaN NaN \n", "ModelID NaN NaN NaN \n", "datasource NaN NaN NaN \n", "auctioneerID NaN NaN NaN \n", "YearMade NaN NaN NaN \n", "MachineHoursCurrentMeter NaN NaN NaN \n", "UsageBand 33985 NaN NaN \n", "saledate 1932 1989-01-17 00:00:00 2011-12-30 00:00:00 \n", "fiModelDesc 5039 NaN NaN \n", "fiBaseModel 19798 NaN NaN \n", "fiSecondaryDesc 43235 NaN NaN \n", "fiModelSeries 13202 NaN NaN \n", "fiModelDescriptor 15875 NaN NaN \n", "ProductSize 62274 NaN NaN \n", "fiProductClassDesc 56166 NaN NaN \n", "state 63944 NaN NaN \n", "ProductGroup 101167 NaN NaN \n", "ProductGroupDesc 101167 NaN NaN \n", "Drive_System 46139 NaN NaN \n", "Enclosure 173932 NaN NaN \n", "Forks 178300 NaN NaN \n", "Pad_Type 70614 NaN NaN \n", "Ride_Control 77685 NaN NaN \n", "Stick 48829 NaN NaN \n", "Transmission 140328 NaN NaN \n", "Turbocharged 75211 NaN NaN \n", "Blade_Extension 24692 NaN NaN \n", "Blade_Width 9615 NaN NaN \n", "Enclosure_Type 21923 NaN NaN \n", "Engine_Horsepower 23937 NaN NaN \n", "Hydraulics 141404 NaN NaN \n", "Pushblock 19463 NaN NaN \n", "Ripper 83452 NaN NaN \n", "Scarifier 12719 NaN NaN \n", "Tip_Control 16207 NaN NaN \n", "Tire_Size 46339 NaN NaN \n", "Coupler 184582 NaN NaN \n", "Coupler_System 40430 NaN NaN \n", "Grouser_Tracks 40515 NaN NaN \n", "Hydraulics_Flow 42784 NaN NaN \n", "Track_Type 84880 NaN NaN \n", "Undercarriage_Pad_Width 79651 NaN NaN \n", "Stick_Length 78820 NaN NaN \n", "Thumb 83093 NaN NaN \n", "Pattern_Changer 90255 NaN NaN \n", "Grouser_Type 84653 NaN NaN \n", "Backhoe_Mounting 78652 NaN NaN \n", "Blade_Type 38612 NaN NaN \n", "Travel_Controls 69923 NaN NaN \n", "Differential_Type 68073 NaN NaN \n", "Steering_Controls 68679 NaN NaN \n", "\n", " mean std min 25% \\\n", "SalesID 1.91971e+06 909021 1.13925e+06 1.41837e+06 \n", "SalePrice 31099.7 23036.9 4750 14500 \n", "MachineID 1.2179e+06 440992 0 1.0887e+06 \n", "ModelID 6889.7 6221.78 28 3259 \n", "datasource 134.666 8.96224 121 132 \n", "auctioneerID 6.55604 16.9768 0 1 \n", "YearMade 1899.16 291.797 1000 1985 \n", "MachineHoursCurrentMeter 3457.96 27590.3 0 0 \n", "UsageBand NaN NaN NaN NaN \n", "saledate NaN NaN NaN NaN \n", "fiModelDesc NaN NaN NaN NaN \n", "fiBaseModel NaN NaN NaN NaN \n", "fiSecondaryDesc NaN NaN NaN NaN \n", "fiModelSeries NaN NaN NaN NaN \n", "fiModelDescriptor NaN NaN NaN NaN \n", "ProductSize NaN NaN NaN NaN \n", "fiProductClassDesc NaN NaN NaN NaN \n", "state NaN NaN NaN NaN \n", "ProductGroup NaN NaN NaN NaN \n", "ProductGroupDesc NaN NaN NaN NaN \n", "Drive_System NaN NaN NaN NaN \n", "Enclosure NaN NaN NaN NaN \n", "Forks NaN NaN NaN NaN \n", "Pad_Type NaN NaN NaN NaN \n", "Ride_Control NaN NaN NaN NaN \n", "Stick NaN NaN NaN NaN \n", "Transmission NaN NaN NaN NaN \n", "Turbocharged NaN NaN NaN NaN \n", "Blade_Extension NaN NaN NaN NaN \n", "Blade_Width NaN NaN NaN NaN \n", "Enclosure_Type NaN NaN NaN NaN \n", "Engine_Horsepower NaN NaN NaN NaN \n", "Hydraulics NaN NaN NaN NaN \n", "Pushblock NaN NaN NaN NaN \n", "Ripper NaN NaN NaN NaN \n", "Scarifier NaN NaN NaN NaN \n", "Tip_Control NaN NaN NaN NaN \n", "Tire_Size NaN NaN NaN NaN \n", "Coupler NaN NaN NaN NaN \n", "Coupler_System NaN NaN NaN NaN \n", "Grouser_Tracks NaN NaN NaN NaN \n", "Hydraulics_Flow NaN NaN NaN NaN \n", "Track_Type NaN NaN NaN NaN \n", "Undercarriage_Pad_Width NaN NaN NaN NaN \n", "Stick_Length NaN NaN NaN NaN \n", "Thumb NaN NaN NaN NaN \n", "Pattern_Changer NaN NaN NaN NaN \n", "Grouser_Type NaN NaN NaN NaN \n", "Backhoe_Mounting NaN NaN NaN NaN \n", "Blade_Type NaN NaN NaN NaN \n", "Travel_Controls NaN NaN NaN NaN \n", "Differential_Type NaN NaN NaN NaN \n", "Steering_Controls NaN NaN NaN NaN \n", "\n", " 50% 75% max \n", "SalesID 1.63942e+06 2.24271e+06 6.33334e+06 \n", "SalePrice 24000 40000 142000 \n", "MachineID 1.27949e+06 1.46807e+06 2.48633e+06 \n", "ModelID 4604 8724 37198 \n", "datasource 132 136 172 \n", "auctioneerID 2 4 99 \n", "YearMade 1995 2000 2013 \n", "MachineHoursCurrentMeter 0 3025 2.4833e+06 \n", "UsageBand NaN NaN NaN \n", "saledate NaN NaN NaN \n", "fiModelDesc NaN NaN NaN \n", "fiBaseModel NaN NaN NaN \n", "fiSecondaryDesc NaN NaN NaN \n", "fiModelSeries NaN NaN NaN \n", "fiModelDescriptor NaN NaN NaN \n", "ProductSize NaN NaN NaN \n", "fiProductClassDesc NaN NaN NaN \n", "state NaN NaN NaN \n", "ProductGroup NaN NaN NaN \n", "ProductGroupDesc NaN NaN NaN \n", "Drive_System NaN NaN NaN \n", "Enclosure NaN NaN NaN \n", "Forks NaN NaN NaN \n", "Pad_Type NaN NaN NaN \n", "Ride_Control NaN NaN NaN \n", "Stick NaN NaN NaN \n", "Transmission NaN NaN NaN \n", "Turbocharged NaN NaN NaN \n", "Blade_Extension NaN NaN NaN \n", "Blade_Width NaN NaN NaN \n", "Enclosure_Type NaN NaN NaN \n", "Engine_Horsepower NaN NaN NaN \n", "Hydraulics NaN NaN NaN \n", "Pushblock NaN NaN NaN \n", "Ripper NaN NaN NaN \n", "Scarifier NaN NaN NaN \n", "Tip_Control NaN NaN NaN \n", "Tire_Size NaN NaN NaN \n", "Coupler NaN NaN NaN \n", "Coupler_System NaN NaN NaN \n", "Grouser_Tracks NaN NaN NaN \n", "Hydraulics_Flow NaN NaN NaN \n", "Track_Type NaN NaN NaN \n", "Undercarriage_Pad_Width NaN NaN NaN \n", "Stick_Length NaN NaN NaN \n", "Thumb NaN NaN NaN \n", "Pattern_Changer NaN NaN NaN \n", "Grouser_Type NaN NaN NaN \n", "Backhoe_Mounting NaN NaN NaN \n", "Blade_Type NaN NaN NaN \n", "Travel_Controls NaN NaN NaN \n", "Differential_Type NaN NaN NaN \n", "Steering_Controls NaN NaN NaN " ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "display_all(df_raw.describe(include='all').T)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "It's important to note what metric is being used for a project. Generally, selecting the metric(s) is an important part of the project setup. However, in this case Kaggle tells us what metric to use: RMSLE (root mean squared log error) between the actual and predicted auction prices. Therefore we take the log of the prices, so that RMSE will give us what we need." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df_raw.SalePrice = np.log(df_raw.SalePrice)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Initial processing" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "ename": "ValueError", "evalue": "could not convert string to float: 'Conventional'", "output_type": "error", "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mValueError\u001b[0m Traceback (most recent call last)", "\u001b[0;32m\u001b[0m in \u001b[0;36m\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[0mm\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mRandomForestRegressor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mn_jobs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 2\u001b[0m \u001b[0;31m# The following code is supposed to fail due to string values in the input data\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mm\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdf_raw\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdrop\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'SalePrice'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maxis\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdf_raw\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSalePrice\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m", "\u001b[0;32m~/anaconda3/lib/python3.6/site-packages/sklearn/ensemble/forest.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, sample_weight)\u001b[0m\n\u001b[1;32m 245\u001b[0m \"\"\"\n\u001b[1;32m 246\u001b[0m \u001b[0;31m# Validate or convert input data\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 247\u001b[0;31m \u001b[0mX\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcheck_array\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maccept_sparse\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m\"csc\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mDTYPE\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 248\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcheck_array\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maccept_sparse\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'csc'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mensure_2d\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 249\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0msample_weight\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", "\u001b[0;32m~/anaconda3/lib/python3.6/site-packages/sklearn/utils/validation.py\u001b[0m in \u001b[0;36mcheck_array\u001b[0;34m(array, accept_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, warn_on_dtype, estimator)\u001b[0m\n\u001b[1;32m 431\u001b[0m force_all_finite)\n\u001b[1;32m 432\u001b[0m \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 433\u001b[0;31m \u001b[0marray\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0morder\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0morder\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcopy\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 434\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 435\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mensure_2d\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", "\u001b[0;31mValueError\u001b[0m: could not convert string to float: 'Conventional'" ] } ], "source": [ "m = RandomForestRegressor(n_jobs=-1)\n", "# The following code is supposed to fail due to string values in the input data\n", "m.fit(df_raw.drop('SalePrice', axis=1), df_raw.SalePrice)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This dataset contains a mix of **continuous** and **categorical** variables.\n", "\n", "The following method extracts particular date fields from a complete datetime for the purpose of constructing categoricals. You should always consider this feature extraction step when working with date-time. Without expanding your date-time into these additional fields, you can't capture any trend/cyclical behavior as a function of time at any of these granularities." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0 2006\n", "1 2004\n", "2 2004\n", "3 2011\n", "4 2009\n", "Name: saleYear, dtype: int64" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "add_datepart(df_raw, 'saledate')\n", "df_raw.saleYear.head()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The categorical variables are currently stored as strings, which is inefficient, and doesn't provide the numeric coding required for a random forest. Therefore we call `train_cats` to convert strings to pandas categories." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "train_cats(df_raw)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can specify the order to use for categorical variables if we wish:" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Index(['High', 'Low', 'Medium'], dtype='object')" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_raw.UsageBand.cat.categories" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df_raw.UsageBand.cat.set_categories(['High', 'Medium', 'Low'], ordered=True, inplace=True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Normally, pandas will continue displaying the text categories, while treating them as numerical data internally. Optionally, we can replace the text categories with numbers, which will make this variable non-categorical, like so:." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df_raw.UsageBand = df_raw.UsageBand.cat.codes" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We're still not quite done - for instance we have lots of missing values, which we can't pass directly to a random forest." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Backhoe_Mounting 0.803872\n", "Blade_Extension 0.937129\n", "Blade_Type 0.800977\n", "Blade_Width 0.937129\n", "Coupler 0.466620\n", "Coupler_System 0.891660\n", "Differential_Type 0.826959\n", "Drive_System 0.739829\n", "Enclosure 0.000810\n", "Enclosure_Type 0.937129\n", "Engine_Horsepower 0.937129\n", "Forks 0.521154\n", "Grouser_Tracks 0.891899\n", "Grouser_Type 0.752813\n", "Hydraulics 0.200823\n", "Hydraulics_Flow 0.891899\n", "MachineHoursCurrentMeter 0.644089\n", "MachineID 0.000000\n", "ModelID 0.000000\n", "Pad_Type 0.802720\n", "Pattern_Changer 0.752651\n", "ProductGroup 0.000000\n", "ProductGroupDesc 0.000000\n", "ProductSize 0.525460\n", "Pushblock 0.937129\n", "Ride_Control 0.629527\n", "Ripper 0.740388\n", "SalePrice 0.000000\n", "SalesID 0.000000\n", "Scarifier 0.937102\n", "Steering_Controls 0.827064\n", "Stick 0.802720\n", "Stick_Length 0.752651\n", "Thumb 0.752476\n", "Tip_Control 0.937129\n", "Tire_Size 0.763869\n", "Track_Type 0.752813\n", "Transmission 0.543210\n", "Travel_Controls 0.800975\n", "Turbocharged 0.802720\n", "Undercarriage_Pad_Width 0.751020\n", "UsageBand 0.000000\n", "YearMade 0.000000\n", "auctioneerID 0.050199\n", "datasource 0.000000\n", "fiBaseModel 0.000000\n", "fiModelDesc 0.000000\n", "fiModelDescriptor 0.820707\n", "fiModelSeries 0.858129\n", "fiProductClassDesc 0.000000\n", "fiSecondaryDesc 0.342016\n", "saleDay 0.000000\n", "saleDayofweek 0.000000\n", "saleDayofyear 0.000000\n", "saleElapsed 0.000000\n", "saleIs_month_end 0.000000\n", "saleIs_month_start 0.000000\n", "saleIs_quarter_end 0.000000\n", "saleIs_quarter_start 0.000000\n", "saleIs_year_end 0.000000\n", "saleIs_year_start 0.000000\n", "saleMonth 0.000000\n", "saleWeek 0.000000\n", "saleYear 0.000000\n", "state 0.000000\n", "dtype: float64" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "display_all(df_raw.isnull().sum().sort_index()/len(df_raw))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "But let's save this file for now, since it's already in format can we be stored and accessed efficiently." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "os.makedirs('tmp', exist_ok=True)\n", "df_raw.to_feather('tmp/bulldozers-raw')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Pre-processing" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the future we can simply read it from this fast format." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df_raw = pd.read_feather('tmp/bulldozers-raw')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We'll replace categories with their numeric codes, handle missing continuous values, and split the dependent variable into a separate variable." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df, y, nas = proc_df(df_raw, 'SalePrice')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We now have something we can pass to a random forest!" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.9830962179413453" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "m = RandomForestRegressor(n_jobs=-1)\n", "m.fit(df, y)\n", "m.score(df,y)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In statistics, the coefficient of determination, denoted R2 or r2 and pronounced \"R squared\", is the proportion of the variance in the dependent variable that is predictable from the independent variable(s). https://en.wikipedia.org/wiki/Coefficient_of_determination" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Wow, an r^2 of 0.98 - that's great, right? Well, perhaps not...\n", "\n", "Possibly **the most important idea** in machine learning is that of having separate training & validation data sets. As motivation, suppose you don't divide up your data, but instead use all of it. And suppose you have lots of parameters:\n", "\n", "\"\"\n", "
\n", "[Underfitting and Overfitting](https://datascience.stackexchange.com/questions/361/when-is-a-model-underfitted)\n", "
\n", "\n", "The error for the pictured data points is lowest for the model on the far right (the blue curve passes through the red points almost perfectly), yet it's not the best choice. Why is that? If you were to gather some new data points, they most likely would not be on that curve in the graph on the right, but would be closer to the curve in the middle graph.\n", "\n", "This illustrates how using all our data can lead to **overfitting**. A validation set helps diagnose this problem." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "((389125, 66), (389125,), (12000, 66))" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "def split_vals(a,n): return a[:n].copy(), a[n:].copy()\n", "\n", "n_valid = 12000 # same as Kaggle's test set size\n", "n_trn = len(df)-n_valid\n", "raw_train, raw_valid = split_vals(df_raw, n_trn)\n", "X_train, X_valid = split_vals(df, n_trn)\n", "y_train, y_valid = split_vals(y, n_trn)\n", "\n", "X_train.shape, y_train.shape, X_valid.shape" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Random Forests" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Base model" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Let's try our model again, this time with separate training and validation sets." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "def rmse(x,y): return math.sqrt(((x-y)**2).mean())\n", "\n", "def print_score(m):\n", " res = [rmse(m.predict(X_train), y_train), rmse(m.predict(X_valid), y_valid),\n", " m.score(X_train, y_train), m.score(X_valid, y_valid)]\n", " if hasattr(m, 'oob_score_'): res.append(m.oob_score_)\n", " print(res)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 1min 4s, sys: 372 ms, total: 1min 4s\n", "Wall time: 8.56 s\n", "[0.0904611534175684, 0.2517003033389636, 0.9828975209204237, 0.8868601882297901]\n" ] } ], "source": [ "m = RandomForestRegressor(n_jobs=-1)\n", "%time m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "An r^2 in the high-80's isn't bad at all (and the RMSLE puts us around rank 100 of 470 on the Kaggle leaderboard), but we can see from the validation set score that we're over-fitting badly. To understand this issue, let's simplify things down to a single small tree." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Speeding things up" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df_trn, y_trn, nas = proc_df(df_raw, 'SalePrice', subset=30000, na_dict=nas)\n", "X_train, _ = split_vals(df_trn, 20000)\n", "y_train, _ = split_vals(y_trn, 20000)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 6.45 s, sys: 132 ms, total: 6.58 s\n", "Wall time: 539 ms\n", "[0.11129876897034846, 0.3501178589366683, 0.9730338701182212, 0.7810845051996299]\n" ] } ], "source": [ "m = RandomForestRegressor(n_jobs=-1)\n", "%time m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Single tree" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.4965829795739235, 0.5246832258551836, 0.50149617735615859, 0.5083655198087873]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=1, max_depth=3, bootstrap=False, n_jobs=-1)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "image/svg+xml": [ "\n", "\n", "\n", "\n", "\n", "\n", "Tree\n", "\n", "\n", "0\n", "\n", "Coupler_System ≤ 0.5\n", "mse = 0.495\n", "samples = 20000\n", "value = 10.189\n", "\n", "\n", "1\n", "\n", "Enclosure ≤ 2.0\n", "mse = 0.414\n", "samples = 16815\n", "value = 10.345\n", "\n", "\n", "0->1\n", "\n", "\n", "True\n", "\n", "\n", "8\n", "\n", "YearMade ≤ 1999.5\n", "mse = 0.109\n", "samples = 3185\n", "value = 9.363\n", "\n", "\n", "0->8\n", "\n", "\n", "False\n", "\n", "\n", "2\n", "\n", "ModelID ≤ 4573.0\n", "mse = 0.331\n", "samples = 4400\n", "value = 9.955\n", "\n", "\n", "1->2\n", "\n", "\n", "\n", "\n", "5\n", "\n", "Enclosure ≤ 4.5\n", "mse = 0.37\n", "samples = 12415\n", "value = 10.484\n", "\n", "\n", "1->5\n", "\n", "\n", "\n", "\n", "3\n", "\n", "mse = 0.315\n", "samples = 2002\n", "value = 10.226\n", "\n", "\n", "2->3\n", "\n", "\n", "\n", "\n", "4\n", "\n", "mse = 0.232\n", "samples = 2398\n", "value = 9.728\n", "\n", "\n", "2->4\n", "\n", "\n", "\n", "\n", "6\n", "\n", "mse = 0.29\n", "samples = 7193\n", "value = 10.736\n", "\n", "\n", "5->6\n", "\n", "\n", "\n", "\n", "7\n", "\n", "mse = 0.272\n", "samples = 5222\n", "value = 10.137\n", "\n", "\n", "5->7\n", "\n", "\n", "\n", "\n", "9\n", "\n", "fiProductClassDesc ≤ 40.5\n", "mse = 0.101\n", "samples = 468\n", "value = 8.988\n", "\n", "\n", "8->9\n", "\n", "\n", "\n", "\n", "12\n", "\n", "saleElapsed ≤ 1217462400.0\n", "mse = 0.082\n", "samples = 2717\n", "value = 9.427\n", "\n", "\n", "8->12\n", "\n", "\n", "\n", "\n", "10\n", "\n", "mse = 0.069\n", "samples = 293\n", "value = 8.896\n", "\n", "\n", "9->10\n", "\n", "\n", "\n", "\n", "11\n", "\n", "mse = 0.118\n", "samples = 175\n", "value = 9.143\n", "\n", "\n", "9->11\n", "\n", "\n", "\n", "\n", "13\n", "\n", "mse = 0.062\n", "samples = 1347\n", "value = 9.529\n", "\n", "\n", "12->13\n", "\n", "\n", "\n", "\n", "14\n", "\n", "mse = 0.081\n", "samples = 1370\n", "value = 9.328\n", "\n", "\n", "12->14\n", "\n", "\n", "\n", "\n", "\n" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "draw_tree(m.estimators_[0], df_trn, precision=3)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Let's see what happens if we create a bigger tree." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[6.526751786450488e-17, 0.38473652894699306, 1.0, 0.73565273648797624]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=1, bootstrap=False, n_jobs=-1)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The training set result looks great! But the validation set is worse than our original model. This is why we need to use *bagging* of multiple trees to get more generalizable results." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Bagging" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Intro to bagging" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "To learn about bagging in random forests, let's start with our basic model again." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.11745691160954547, 0.27959279688230376, 0.97139456205050101, 0.86039533492251219]\n" ] } ], "source": [ "m = RandomForestRegressor(n_jobs=-1)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We'll grab the predictions for each individual tree, and look at one example." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(array([ 9.21034, 8.9872 , 8.9872 , 8.9872 , 8.9872 , 9.21034, 8.92266, 9.21034, 9.21034, 8.9872 ]),\n", " 9.0700003890739005,\n", " 9.1049798563183568)" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "preds = np.stack([t.predict(X_valid) for t in m.estimators_])\n", "preds[:,0], np.mean(preds[:,0]), y_valid[0]" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(10, 12000)" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "preds.shape" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD8CAYAAACb4nSYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot([metrics.r2_score(y_valid, np.mean(preds[:i+1], axis=0)) for i in range(10)]);" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The shape of this curve suggests that adding more trees isn't going to help us much. Let's check. (Compare this to our original model on a sample)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.10721195540628872, 0.2777358026154778, 0.9761670456844791, 0.86224362387001874]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=20, n_jobs=-1)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.1029319603663909, 0.2725488716109634, 0.97803192843821529, 0.86734099039701873]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=40, n_jobs=-1)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.09942284423261978, 0.27026457977935875, 0.97950425012208453, 0.86955536025947799]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=80, n_jobs=-1)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Out-of-bag (OOB) score" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Is our validation set worse than our training set because we're over-fitting, or because the validation set is for a different time period, or a bit of both? With the existing information we've shown, we can't tell. However, random forests have a very clever trick called *out-of-bag (OOB) error* which can handle this (and more!)\n", "\n", "The idea is to calculate error on the training set, but only include the trees in the calculation of a row's error where that row was *not* included in training that tree. This allows us to see whether the model is over-fitting, without needing a separate validation set.\n", "\n", "This also has the benefit of allowing us to see whether our model generalizes, even if we only have a small amount of data so want to avoid separating some out to create a validation set.\n", "\n", "This is as simple as adding one more parameter to our model constructor. We print the OOB error last in our `print_score` function below." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.10198464613020647, 0.2714485881623037, 0.9786192457999483, 0.86840992079038759, 0.84831537630038534]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=40, n_jobs=-1, oob_score=True)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This shows that our validation set time difference is making an impact, as is model over-fitting." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Reducing over-fitting" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Subsampling" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "It turns out that one of the easiest ways to avoid over-fitting is also one of the best ways to speed up analysis: *subsampling*. Let's return to using our full dataset, so that we can demonstrate the impact of this technique." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df_trn, y_trn, nas = proc_df(df_raw, 'SalePrice')\n", "X_train, X_valid = split_vals(df_trn, n_trn)\n", "y_train, y_valid = split_vals(y_trn, n_trn)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The basic idea is this: rather than limit the total amount of data that our model can access, let's instead limit it to a *different* random subset per tree. That way, given enough trees, the model can still see *all* the data, but for each individual tree it'll be just as fast as if we had cut down our dataset as before." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "set_rf_samples(20000)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 8.38 s, sys: 428 ms, total: 8.81 s\n", "Wall time: 3.49 s\n", "[0.24021020451254516, 0.2780622994610262, 0.87937208432405256, 0.86191954999425424, 0.86692047674867767]\n" ] } ], "source": [ "m = RandomForestRegressor(n_jobs=-1, oob_score=True)\n", "%time m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Since each additional tree allows the model to see more data, this approach can make additional trees more useful." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.2317315086850927, 0.26334275954117264, 0.89225792718146846, 0.87615150359885019, 0.88097587673696554]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=40, n_jobs=-1, oob_score=True)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Tree building parameters" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We revert to using a full bootstrap sample in order to show the impact of other over-fitting avoidance methods." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "reset_rf_samples()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Let's get a baseline for this full set to compare to." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "def dectree_max_depth(tree):\n", " children_left = tree.children_left\n", " children_right = tree.children_right\n", "\n", " def walk(node_id):\n", " if (children_left[node_id] != children_right[node_id]):\n", " left_max = 1 + walk(children_left[node_id])\n", " right_max = 1 + walk(children_right[node_id])\n", " return max(left_max, right_max)\n", " else: # leaf\n", " return 1\n", "\n", " root_node_id = 0\n", " return walk(root_node_id)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.07828713008286803, 0.23818558990341943, 0.9871909898049919, 0.8986837887808402, 0.9085077721150765]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=40, n_jobs=-1, oob_score=True)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "t=m.estimators_[0].tree_" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "45" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "dectree_max_depth(t)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.14073508031497292, 0.23337403295759937, 0.9586057939941005, 0.9027357960501001, 0.9068706269691232]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=40, min_samples_leaf=5, n_jobs=-1, oob_score=True)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "t=m.estimators_[0].tree_" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "35" ] }, "execution_count": null, "metadata": {}, "output_type": "execute_result" } ], "source": [ "dectree_max_depth(t)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Another way to reduce over-fitting is to grow our trees less deeply. We do this by specifying (with `min_samples_leaf`) that we require some minimum number of rows in every leaf node. This has two benefits:\n", "\n", "- There are less decision rules for each leaf node; simpler models should generalize better\n", "- The predictions are made by averaging more rows in the leaf node, resulting in less volatility" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.11595869956476182, 0.23427349924625201, 0.97209195463880227, 0.90198460308551043, 0.90843297242839738]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=40, min_samples_leaf=3, n_jobs=-1, oob_score=True)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can also increase the amount of variation amongst the trees by not only use a sample of rows for each tree, but to also using a sample of *columns* for each *split*. We do this by specifying `max_features`, which is the proportion of features to randomly select from at each split." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "- None\n", "- 0.5\n", "- 'sqrt'" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "- 1, 3, 5, 10, 25, 100" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[0.11926975747908228, 0.22869111042050522, 0.97026995966445684, 0.9066000722129437, 0.91144914977164715]\n" ] } ], "source": [ "m = RandomForestRegressor(n_estimators=40, min_samples_leaf=3, max_features=0.5, n_jobs=-1, oob_score=True)\n", "m.fit(X_train, y_train)\n", "print_score(m)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can't compare our results directly with the Kaggle competition, since it used a different validation set (and we can no longer to submit to this competition) - but we can at least see that we're getting similar results to the winners based on the dataset we have.\n", "\n", "The sklearn docs [show an example](http://scikit-learn.org/stable/auto_examples/ensemble/plot_ensemble_oob.html) of different `max_features` methods with increasing numbers of trees - as you see, using a subset of features on each split requires using more trees, but results in better models:\n", "![sklearn max_features chart](http://scikit-learn.org/stable/_images/sphx_glr_plot_ensemble_oob_001.png)" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" } }, "nbformat": 4, "nbformat_minor": 2 }