Abstract

This project will examine the use of random forests in machine intelligence to be able to predict the outcomes of Supreme Court cases, with at least a 70 percent level of proficiency. Random Forests have been used previously for various classification problems. In the medical field Random Forests have been used to predict the risk of a person having a stroke. They have also been used to predict from three different plant species based on various structural features. Importantly, in terms of Judicial behavior, Random Forests have been able to show the correlations between certain Supreme Court case factors that could influence whether Justices will affirm or reverse a case. To evaluate possible correlations and predict Justice's behavior forty years of data (1973-2013) will be evaluated. Found online at the Supreme Court Database, this information should allow the software to correctly predict whether the Supreme Court will affirm or reverse a case. Specific variables will be evaluated from the Supreme Court Database such as the ideological leaning of the Supreme Court at that point in time, the District Court the case was originally heard in, and whether the case was a violent crime or concerned individual rights such as abortion. Programming is accomplished using python and the Scikit-Learn library to develop the random forest classifier for over 40 years of data. To improve classification, Random Forests use an ensemble method that incorporates various machine learning algorithms.

Advisor

Montelione, Thomas

Department

Computer Science

Disciplines

Other Computer Sciences | Theory and Algorithms

Publication Date

2022

Degree Granted

Bachelor of Arts

Document Type

Senior Independent Study Thesis

Share

COinS
 

© Copyright 2022 Chris Savaglio