Implementasi Algoritma LightGBM untuk Prediksi Status Gizi Bayi dan Balita di Desa Doko Kabupaten Kediri

Thoriqulhaq, Muhammad
Idhom, Mohammad
Maulida Hindrayani, Kartika

Jurnal Teknik Terapan (J-TETA)
Vol. 4 No. 2 (2025): Oktober

MThe issue of nutritional status among infants and toddlers remains a serious concern in Indonesia, particularly in rural areas. Doko Village was chosen as the research location due to its significant challenges in child health. This study aims to develop a nutritional status prediction model based on the LightGBM algorithm, capable of processing anthropometric data to classify nutritional categories such as "Underweight", "Normal", and "Overweight". Using an 80:20 training-to-testing data ratio, the model achieved 97% accuracy and a 94% F1-score. In addition to building the prediction model, this study also developed an interactive web application using Streamlit, and compared its results with the conventional WHO AnthroPlus method. The results indicate that LightGBM offers advantages in terms of speed, flexibility, and predictive accuracy based on local data.