Analisis Regresi Logistik Biner dan Random Forest untuk Prediksi Faktor-Faktor Stunting di Pulau Jawa

Abstract

This study aimed to compare the performance and variable identification capabilities of Binary Logistic Regression and Random Forest models in classification analysis. The results showed that both methods consistently identified variables X1, X3, and X4 as the most influential factors in predicting outcomes. However, Binary Logistic Regression also identified variable X6 as statistically significant, which was not reflected in the Random Forest model. In terms of model performance, Random Forest outperformed Binary Logistic Regression across all evaluation metrics, including accuracy, precision, sensitivity, specificity, and balanced accuracy. These findings suggested that Random Forest was more effective in handling complex data structures and delivering optimal classification results, while Binary Logistic Regression excelled in providing deeper interpretability of variable relationships. Therefore, the choice of method should have aligned with the analytical objectives, and combining both approaches could have yielded more comprehensive insights.