PERFORMANCE EVALUATION OF MACHINE LEARNING ALGORITHMS FOR JUDICIAL PREDICTION SYSTEM

Abstract

Artificial Intelligence and Machine Learning techniques have been productively utilized to forecast judicial outcomes and analyze it. A Judicial Prediction System (JPS) is for forecasting the judicial results based on historical data, legal precedents, and other relevant factors, thereby providing judges and other law professionals the predictive insights into case outcomes. The purpose of the JPS is to educate the public by promoting transparency in the legal process and overcoming various factors negatively influencing the final judgment such as cognitive biases, judicial bottlenecks, emotions, and so on. This paper aims to find the most effective way for judicial outcome prediction to assist in time and judicial resource optimization. Four distinct algorithms: Support Vector Machine, Random Forest, Logistic Regression and K-Nearest Neighbor have been utilized to determine the appeal case outcomes at the Supreme Court of Nigeria (SCN). The dataset used in training the machine learning algorithms was obtained locally from a Supreme Court of Nigeria (SCN). The models were evaluated using precision, recall F1 score and accuracy. Results show that Random Forest provided the highest accuracy of 72%. However, future research should consider ensemble approach for judicial cases prediction.

 KEYWORDS: judicial prediction system, Machine Learning algorithms, model, case outcomes