{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Training a Machine Learning model with scikit-learn ([video #4](https://www.youtube.com/watch?v=RlQuVL6-qe8&list=PL5-da3qGB5ICeMbQuqbbCOQWcS6OYBr5A&index=4))\n",
"\n",
"Created by [Data School](https://www.dataschool.io). Watch all 10 videos on [YouTube](https://www.youtube.com/playlist?list=PL5-da3qGB5ICeMbQuqbbCOQWcS6OYBr5A). Download the notebooks from [GitHub](https://github.com/justmarkham/scikit-learn-videos).\n",
"\n",
"**Note:** This notebook uses Python 3.9.1 and scikit-learn 0.23.2. The original notebook (shown in the video) used Python 2.7 and scikit-learn 0.16."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Agenda\n",
"\n",
"- What is the **K-nearest neighbors** classification model?\n",
"- What are the four steps for **model training and prediction** in scikit-learn?\n",
"- How can I apply this pattern to **other Machine Learning models**?"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Reviewing the iris dataset"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"# added empty cell so that the cell numbering matches the video"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"\n",
" \n",
" "
],
"text/plain": [
""
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"from IPython.display import IFrame\n",
"IFrame('https://www.dataschool.io/files/iris.txt', width=300, height=200)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"- 150 **observations**\n",
"- 4 **features** (sepal length, sepal width, petal length, petal width)\n",
"- **Response** variable is the iris species\n",
"- **Classification** problem since response is categorical\n",
"- More information in the [UCI Machine Learning Repository](http://archive.ics.uci.edu/ml/datasets/Iris)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## K-nearest neighbors (KNN) classification"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"1. Pick a value for K.\n",
"2. Search for the K observations in the training data that are \"nearest\" to the measurements of the unknown iris.\n",
"3. Use the most popular response value from the K nearest neighbors as the predicted response value for the unknown iris."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Example training data\n",
"\n",
"![Training data](images/04_knn_dataset.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### KNN classification map (K=1)\n",
"\n",
"![1NN classification map](images/04_1nn_map.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### KNN classification map (K=5)\n",
"\n",
"![5NN classification map](images/04_5nn_map.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"*Image Credits: [Data3classes](https://commons.wikimedia.org/wiki/File:Data3classes.png#/media/File:Data3classes.png), [Map1NN](https://commons.wikimedia.org/wiki/File:Map1NN.png#/media/File:Map1NN.png), [Map5NN](https://commons.wikimedia.org/wiki/File:Map5NN.png#/media/File:Map5NN.png) by Agor153. Licensed under CC BY-SA 3.0*"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Loading the data"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
"# import load_iris function from datasets module\n",
"from sklearn.datasets import load_iris\n",
"\n",
"# save \"bunch\" object containing iris dataset and its attributes\n",
"iris = load_iris()\n",
"\n",
"# store feature matrix in \"X\"\n",
"X = iris.data\n",
"\n",
"# store response vector in \"y\"\n",
"y = iris.target"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"(150, 4)\n",
"(150,)\n"
]
}
],
"source": [
"# print the shapes of X and y\n",
"print(X.shape)\n",
"print(y.shape)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## scikit-learn 4-step modeling pattern"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"**Step 1:** Import the class you plan to use"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [],
"source": [
"from sklearn.neighbors import KNeighborsClassifier"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"**Step 2:** \"Instantiate\" the \"estimator\"\n",
"\n",
"- \"Estimator\" is scikit-learn's term for model\n",
"- \"Instantiate\" means \"make an instance of\""
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [],
"source": [
"knn = KNeighborsClassifier(n_neighbors=1)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"- Name of the object does not matter\n",
"- Can specify tuning parameters (aka \"hyperparameters\") during this step\n",
"- All parameters not specified are set to their defaults"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"KNeighborsClassifier(n_neighbors=1)\n"
]
}
],
"source": [
"print(knn)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"**Step 3:** Fit the model with data (aka \"model training\")\n",
"\n",
"- Model is learning the relationship between X and y\n",
"- Occurs in-place"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"KNeighborsClassifier(n_neighbors=1)"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"knn.fit(X, y)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"**Step 4:** Predict the response for a new observation\n",
"\n",
"- New observations are called \"out-of-sample\" data\n",
"- Uses the information it learned during the model training process"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([2])"
]
},
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"knn.predict([[3, 5, 4, 2]])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"- Returns a NumPy array\n",
"- Can predict for multiple observations at once"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([2, 1])"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"X_new = [[3, 5, 4, 2], [5, 4, 3, 2]]\n",
"knn.predict(X_new)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Using a different value for K"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([1, 1])"
]
},
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# instantiate the model (using the value K=5)\n",
"knn = KNeighborsClassifier(n_neighbors=5)\n",
"\n",
"# fit the model with data\n",
"knn.fit(X, y)\n",
"\n",
"# predict the response for new observations\n",
"knn.predict(X_new)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Using a different classification model"
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([2, 0])"
]
},
"execution_count": 12,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# import the class\n",
"from sklearn.linear_model import LogisticRegression\n",
"\n",
"# instantiate the model\n",
"logreg = LogisticRegression(solver='liblinear')\n",
"\n",
"# fit the model with data\n",
"logreg.fit(X, y)\n",
"\n",
"# predict the response for new observations\n",
"logreg.predict(X_new)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Resources\n",
"\n",
"- [Nearest Neighbors](https://scikit-learn.org/stable/modules/neighbors.html) (user guide), [KNeighborsClassifier](https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.KNeighborsClassifier.html) (class documentation)\n",
"- [Logistic Regression](https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression) (user guide), [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) (class documentation)\n",
"- [Videos from An Introduction to Statistical Learning](https://www.dataschool.io/15-hours-of-expert-machine-learning-videos/)\n",
" - Classification Problems and K-Nearest Neighbors (Chapter 2)\n",
" - Introduction to Classification (Chapter 4)\n",
" - Logistic Regression and Maximum Likelihood (Chapter 4)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Comments or Questions?\n",
"\n",
"- Email: \n",
"- Website: https://www.dataschool.io\n",
"- Twitter: [@justmarkham](https://twitter.com/justmarkham)\n",
"\n",
"© 2021 [Data School](https://www.dataschool.io). All rights reserved."
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.9.4"
}
},
"nbformat": 4,
"nbformat_minor": 1
}