Perbandingan Algoritma Naïve Bayes, Decision Tree, KNN, dan Random Forest Untuk Memprediksi Data Penduduk Penerima BPJS Di Lampung Timur

Abstract

This study aims to predict population data in Lampung Timur using various classification algorithms. The algorithms used include Naive Bayes, k-Nearest Neighbors (k-NN), Decision Tree, and Random Forest. The dataset used was derived from population data processed with RapidMiner. The data was processed using steps such as reading from Excel files, data duplication, and model training with the aforementioned algorithms. Evaluation results show that the Naive Bayes algorithm has the highest accuracy of 86.89% with good precision and recall for both BPJS and UMUM classes. Additional analysis indicates that from the dataset used, there are 1924 residents who have BPJS and 1960 residents who do not have BPJS. These results suggest that the Naive Bayes algorithm performs best in predicting population data in Lampung Timur and that there is still a significant number of residents who do not utilize BPJS services. Implementing this classification algorithm can aid in better decision-making regarding the distribution of BPJS services in Lampung Timur.