{ "nbformat": 4, "nbformat_minor": 0, "metadata": { "colab": { "name": "Decision trees in machine learning.ipynb", "provenance": [], "authorship_tag": "ABX9TyMJ2NJx1qBt04nDfa20LXnj" }, "kernelspec": { "name": "python3", "display_name": "Python 3" }, "language_info": { "name": "python" } }, "cells": [ { "cell_type": "markdown", "source": [ "# Decision trees in machine learning\n", "\n", "## Try me\n", " [![Open In Colab](../../_static/colabs_badge.png)](https://colab.research.google.com/github/ffraile/operations-research-notebooks/blob/main/docs/source/Decision%20Theory/tutorials/Decision_trees_in_machine_learning.ipynb)[![Binder](../../_static/binder_badge.png)](https://mybinder.org/v2/gh/ffraile/operations-research-notebooks/main?labpath=docs%2Fsource%2FDecision%20Theory%2Ftutorials%2FDecision_trees_in_machine_learning.ipynb)\n", "\n", "## Introduction\n", "So far, we have seen how decision trees can be used as tools to support decision-making in complex decision problems. However, decision trees provide the basis for a family of machine learning techniques. The idea behind these techniques is to learn from training data a decision tree that can be used to solve a machine learning tasks. Normally, decision trees are applied in **classification problems** (recall that these problems the objective is to classify input data into known categories based on the features available in the training data).\n", "\n", "The objective is to generate a decision tree which represents the decisions that the machine will take to classify a data sample based on its features. For instance, Let's generate some synthetic data representing sensor data from two sensors, temperature and encoding speed, in two different operating states, normal operation and failure. The following script generates some synthetic data that we can use to guide the example:" ], "metadata": { "id": "OgdyUcu74-5o" } }, { "cell_type": "code", "source": [ "from sklearn.datasets import make_blobs\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "n_samples = 50000\n", "centers = [(25, 25), (33, 32)]\n", "X, y = make_blobs(n_samples=n_samples, centers=centers, shuffle=False, random_state=42)\n", "plt.figure()\n", "y_unique = np.unique(y)\n", "colors = ['blue', 'red']\n", "classes = ['Normal operating state', 'Failure state']\n", "for this_y, color in zip(y_unique, colors):\n", " this_X = X[y == this_y]\n", " \n", " plt.scatter(\n", " this_X[:, 0],\n", " this_X[:, 1],\n", " c=color,\n", " alpha=0.5,\n", " edgecolor=\"k\",\n", " label=classes[this_y],\n", " )\n", "plt.legend(loc=\"best\")\n", "plt.ylabel('rpms/1000')\n", "plt.xlabel('Temperature in degrees')\n", "plt.title(\"Sensor readings\")" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 312 }, "id": "xtIoJ8SZkMz-", "outputId": "7e768794-8945-4f34-db89-5047800998a9" }, "execution_count": null, "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "Text(0.5, 1.0, 'Sensor readings')" ] }, "metadata": {}, "execution_count": 10 }, { "output_type": "display_data", "data": { "text/plain": [ "
" ], "image/png": "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\n" }, "metadata": { "needs_background": "light" } } ] }, { "cell_type": "markdown", "source": [ "Analyzing the data, we can see that the machine is in normal operating state if the rpms are below 29000 rpms and the temperature below 29 degrees Celsius.\n", "The decision tree below shows how these decisions perform the classification.\n", "\n", "![condition monitoring decision tree](img/ml_decision_tree-condition-monitoring.drawio.png)\n", "\n", "As you can see in the figure, we first check the temperature reading and compare it to a threshold value, and it will take one decision branch or the other depending on whether the reading is greater or equal to the threshold. Then, in a subsequent decision node, it will check the readings of the encoder and depending on the value, and the branch, determine if the machine is in normal operating state or in failure state.\n", "\n", "\n", "The script below represents the data and the resulting decision regions: \n" ], "metadata": { "id": "8XM20dxLkNJN" } }, { "cell_type": "code", "source": [ "plt.figure()\n", "\n", "for this_y, color in zip(y_unique, colors):\n", " this_X = X[y == this_y]\n", " \n", " plt.scatter(\n", " this_X[:, 0],\n", " this_X[:, 1],\n", " c=color,\n", " alpha=0.5,\n", " edgecolor=\"k\",\n", " label=\"Class %s\" % this_y,\n", " )\n", "\n", "x = np.linspace(20, 37, 1000)\n", "sep1 = 0*x + 29\n", "sep2 = 0*x + 20\n", "plt.fill_between(x, sep1, sep2, where=x<=30, alpha=0.1, color=colors[0])\n", "sep3 = 0*x + 37\n", "plt.fill_between(x, sep3, sep1, where=x<=30, alpha=0.1, color=colors[1])\n", "plt.fill_between(x, sep3, sep2, where=x>30, alpha=0.1, color=colors[1])\n", "plt.legend(loc=\"best\")\n", "plt.ylabel('rpms/1000')\n", "plt.xlabel('Temperature in degrees')\n", "plt.title(\"Sensor readings\")\n" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 312 }, "id": "WZKANBUGpAo1", "outputId": "3ac1807d-ddcb-4837-fdbd-df2a54943563" }, "execution_count": null, "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "Text(0.5, 1.0, 'Sensor readings')" ] }, "metadata": {}, "execution_count": 11 }, { "output_type": "display_data", "data": { "text/plain": [ "
" ], "image/png": "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\n" }, "metadata": { "needs_background": "light" } } ] }, { "cell_type": "markdown", "source": [ "Now, the objective is to develop an algorithm that is able to learn this decision tree from training data, that is, use the training data to decide which decision nodes are needed at each stage and what are the values of the thresholds that need to be applied. The result is a decision tree that can be used to classify new data, with the following generic structure:\n", "\n", "![ml decision tree](img/ml_decision_tree.png)\n", "\n", "The following section describes the most common strategy used in decision trees algorithms to obtain these decision trees." ], "metadata": { "id": "vjIybaAUpBUT" } }, { "cell_type": "markdown", "source": [ "## Decision Tree Learning Algorithms\n", "### Strategy: Recursive Binary Splitting\n", "The most common strategy used to build decision trees from training data consists in applying a **greedy approach** to split the dataset into subsets, step by step, choosing at every step the feature that *best splits* the data into two subsets, until each subset contains only data samples from one class, or until the subsets are *homogeneous* or *pure* enough, or in other words, the level of **entropy** of the data is low enough.\n", "Moreover, the strategy is to iteratively select one feature to split the data until:\n", "\n", "- All (or almost all) data samples in a leaf node belong to the same class (minimum impurity or minimum entropy).\n", "- The tree has reached the maximum pre-defined number of splits (maximum depth).\n", "- The tree has tried all features and found that none of them is useful to split the data anymore.\n", "\n", "There are some concepts in the paragraph above that we need to unpack, mainly related to the concepts of data homogeneity and entropy.\n", "\n", "#### Data Homogeneity\n", "Data homogeneity is the condition that all data samples in a leaf node belong to the same class. The **Gini** impurity measure is a measure of data homogeneity in a dataset $S$), expressed as:\n", "\n", "$\\text{Gini}(S) = 1 - \\sum_{i=1}^c{p_i^2}$\n", "\n", "Where $C={1, 2, ..., c}$ is the set of possible classes, and $p_i$ is the probability of class $i$, defined as the number of data samples of class $i$ divided by the total number of data samples in the dataset. From the definition above, we can see that the impurity is 0 if all the data samples in the dataset are of the same class.\n", "\n", "#### Entropy\n", "In information theory, the entropy is a measure of the level of uncertainty in a system. Formally, the entropy of a system is the sum of the probability of each possible state of the system multiplied by the logarithm of the probability of each possible state of the system:\n", "\n", "$\\text{Entropy}(S) = \\sum_{i=1}^c p_i \\log_2 p_i$\n", "\n", "To understand this measure, consider tossing a coin. If the coin is fair, the probability of getting heads is equal to the probability of getting tails, and the entropy of the system is 1, representing total disorder, or chaos, because the uncertainty of the result of the toss is maximum. However, if the coin is completely biased, for instance, the probability of getting heads is 1, then the entropy is 0, because the result is certain, and completely predictable.\n", "\n", "We can see that the measure of entropy and the measure of impurity are equivalent, in the sense that we can use the entropy to assess data homogeneity in a data subset.\n", "\n", "#### Recursive partitioning\n", "As mentioned above, the tree is built by partitioning the dataset into two subsets, selecting in each iteration the feature and the threshold that maximizes data homogeneity (minimizes entropy or impurity). The partition is **recursive** because it is applied to every node of the tree (that is, the first partition results into two subsets, and then the partition is applied again to each subset, resulting in four subsets, and so forth).\n", "\n", "At every iteration, the data at any node $m$ is partitioned into two datasets, and the data Homogeneity of the two datasets is evaluated for every candidate split $\\theta$, expressed as two decision variables, the feature $j$ and a threshold $t_m$: $\\theta = (j, t_m)$. Then, for instance, gradient descent methods can be used to find the optimal $\\theta$ for the current node $m$ that minimizes the data homogeneity.\n", "\n", "Formally, let the data subset at node $m$ be $S_m$, consisting of $N_m$ data samples, each containing a feature vector $x$ and a label $y$. The candidate $\\theta = (j, t_m)$ partition the data into two subsets $S_m^{left}(\\theta)$ such that:\n", "\n", "\n", "$S_m^{left}(\\theta) = {x \\in S_m: x_j \\leq t_m}$\n", "\n", "And and $S_m^{right}(\\theta)$ such that:\n", "\n", "$S_m^{right}(\\theta) = {x \\in S_m: x_j > t_m}$\n", "\n", "Let us now define the loss function for the current node $m$:\n", "\n", "$\\text{L}(S_m,\\theta) = \\frac{N_m^{left}}{N_m}\\text{H}(S_m^{left}(\\theta)) + \\frac{N_m^{right}}{N_m}\\text{H}(S_m^{right}(\\theta))$\n", "\n", "where $N_m^{left}$ and $N_m^{right}$ are the number of data samples in the left and right subsets, respectively, and $\\text{H}(S)$ is a measure of the data entropy or impurity in the dataset $S$.\n", "\n", "This way, the learning problem is reduced to a minimization of the loss function, which is equivalent to finding the $\\theta$ that minimizes the data Homogeneity.\n", "\n", "##### Information gain\n", "Another approach is to maximise the information gain, which is the difference between the data Homogeneity of the data subset at node $m$ and the data homogeneity of the two resulting subsets.\n", "\n", "$\\text{I}(S_m) = \\text{H}(S_m)-\\frac{N_m^{left}}{N_m}\\text{H}(S_m^{left}) - \\frac{N_m^{right}}{N_m}\\text{H}(S_m^{right})$\n", "\n", "\n", "That is, the objective now is to find the $\\theta$ that maximizes the difference between the data homogeneity before and after the split.\n", "\n", "#### Algorithms\n", "There are different algorithms that use this approach to build decision trees, the most popular being Iterative Dichotomiser 3 (ID3), its successors C4.5 and C5.0., and finally the Classification and Regression Trees (CART) algorithm.\n", "\n", "ID3 uses the greedy recursive partitioning algorithm finding for each node the feature and threshold that maximizes the information gain, with the limitation that the features must be categorical or qualitative (that is, it can only take one of a finite number of values). C4.5 and C5.0 on the other hand remove this restriction. CART is very similar to C4.5, but it provides additional support to solve regression problems.\n", "\n", "\n", "#### Pruning\n", "All the algorithm above have a common drawback: They tend to overfitting the training data. To overcome this, they use a pruning strategy that removes the nodes that do not improve the model performance. One common pruning strategy is cost complexity pruning, where the nodes are removed (pruned) if they do not improve the model performance. Minimum cost complexity pruning finds the best trade-off between the tree complexity (expressed as the number of terminal nodes) and performance.\n", "\n" ], "metadata": { "collapsed": false } } ] }