Classification of Remission Data for Prisoners in Tangerang Class IIA Correctional Institution using K-Nearest Neighbor Algorithm

Abstract

The classification of remission eligibility for prisoners is a critical issue in correctional institutions, as it directly impacts prison management and the rehabilitation process. Special remission is a reduction of sentence granted to prisoners based on specific criteria, including religious status and the type of remission granted. This research aims to address the challenge of classifying special remission data for prisoners at the Class IIA Tangerang Correctional Facility using the K-Nearest Neighbor (KNN) algorithm. The dataset used in this study includes four indicators: Length of Sentence, Remaining Sentence, Crime Type, and Risk Dimension, which are analyzed to predict the remission status to be granted. The KNN model, with a parameter of k=1, achieved an accuracy of 93.94%. However, the model struggled to accurately classify the "No Remission" class, resulting in failures to detect prisoners who are not eligible for remission. The data processing steps included converting categorical data into numerical format, data normalization, and splitting the data into training and testing sets. Model evaluation was conducted using Confusion Matrix, Precision, Recall, and F1-Score. The findings suggest that while the KNN algorithm can be effectively used to classify remission status, further improvements are needed to address class imbalance and optimize results.