Analysis of Data Mining in Predicting Poverty Levels in Indonesia Using the Decision Tree Method

Analisa Data Mining Dalam Memprediksi Tingkat Kemiskinan Masyarakat Indonesia Dengan Metode Decision Tree

Abstract

This study aims to examine the application of the Decision Tree method in predicting poverty levels in Indonesia using the RapidMiner software. Poverty is a complex issue influenced by social, economic, and educational factors. Through a data mining approach, this research seeks to identify patterns within poverty data to support more accurate decision-making. The research data were obtained from the public platform Kaggle and include key variables such as individual expenditure, the Human Development Index (HDI), average study time, access to proper sanitation and safe drinking water, as well as the open unemployment rate. The results show that the Decision Tree model achieved an accuracy of 94.90%, with a precision of 95.24% and a recall of 93.75%, based on the confusion matrix. The use of RapidMiner also facilitates the analysis, as the results are presented visually and are easy to understand. This model is recommended for implementation in government information