Pemanfaatan Algoritma Levenberg-Marquardt untuk Analisis Prediksi Persentase Penduduk yang Melakukan Pengobatan Sendiri

Self-medication is a practice in which individuals use drugs or administer treatments without a doctor's prescription or medical supervision. This phenomenon has become a significant health issue in Indonesia, particularly in the city of Pematangsiantar and Simalungun Regency, where many residents tend to self-medicate without receiving adequate medical consultation. Therefore, the aim of this study is to analyze the predicted percentage of health independence development among residents who self-medicate in Pematangsiantar and Simalungun Regency using the Levenberg-Marquardt algorithm. The research data consists of time-series data on the percentage of residents self-medicating in Pematangsiantar and Simalungun Regency from 2018 to 2023, obtained from the Central Statistics Agency of North Sumatra. The analysis was conducted using five architecture models: 4-5-1, 4-10-1, 4-15-1, 4-20-1, and 4-25-1. The results show that the Levenberg-Marquardt algorithm with the 4-15-1 architecture model provided the best performance, with the lowest Mean Squared Error (MSE) value of 0.0268691174 compared to the other architecture models. This study is expected to assist local governments by providing information on the development of the percentage of residents who self-medicate in Pematangsiantar and Simalungun Regency, enabling them to formulate the best policies for improving public health in the region in the future. This research also contributes to the development of artificial intelligence-based health prediction methods, particularly for analyzing the percentage of self-medicating residents in complex and dynamic regional contexts.