RECIDIVISM

In this next section, we explore two methods, Classification Trees and Random Forests in trying to predict recidivism. First, let's take a look at the data.

Description of the Data

The data being used consists of 10,428 criminals in Broward County, Florida from the years 2013 to 2014 and contains past criminal history and what they were charged for when they were arrested. Moreover, it contains the outcome (recidivated / did not recidivate) after two years of receiving their COMPASS risk assessment. For a description of each variable, please click here.

Classification Trees

Our first method of machine learning that we will look at are classification trees. To familiarize yourself, please watch the video below that goes through the process of creating a classification tree by using a sample of the data we will be using.

With measuring accuracies, it is also important to measure the accuracy of a model as it relates to different groups. This is a good way to check if the model may be inherently biased towards a specific group.

Investigating Your Own Classification Tree

Use the app below to make your own Classification Trees. The app allows you to choose which variables go into making a classification tree and how bushy (# number of branches) your tree can be. Also, you can see how well your tree performs by looking at the confusion matrix and overall accuracy. You can also view the accuracy of your tree by race, sex, marital status, and age category.

To use the app below, start with the following settings then answer the questions to the left.

• Past History of the Criminal: priors
• Charactericstics of the Criminal: age sex
• Type of Crime Committed: Null
• See accuracy accross: none

Predicting if someone will recidivate is a difficult task. While decision trees are powerful classifying algorithms, they have their advantages and disadvantages. Unlike many machine learning algorithms, decision trees are white-box, meaning the user can see the steps taken in order to classify an individual, and the construction of the tree can be clearly explained; furthermore, they are easy to understand and interpret. Generally, though, decision trees alone are not useful as they are unstable, lack flexibility, and are relatively inaccurate compared to other algorithms.

While the COMPAS algorithm is proprietary, its mechanisms are likely more complex than a standard decision tree. By adopting concepts from decision trees, we can build a more advanced model known as a random forest. Instead of outputting one specific category, a random forest returns a probability of some outcome occurring based on the results of many decision trees. For example, the decision tree for recisivism in the above app will result in either a Yes or No response. Random forest algorithms provide the probability of a person recidivating. When used to predict recidivism, we can use this probabilistic output to create a more equitable model.

A very useful aspect of random forests is that the result is an aggregation of many different, uncorrelated decision trees. Each tree in a random forest is made from bootstrapped data -- a method where a new sample is created by randomly selecting rows from the original sample (with replacement). Another reason random forests produce different trees is because only a subset of variables are considered at each node of a tree’s construction. Since trees are created with bootstrapping and we aggregate the results of many trees to return a probability of an outcome, a random forest is known as a bagging algorithm.