Penerapan Metode Logistic Regression dalam Sistem Prediksi Risiko Stunting Anak Berbasis Web

Rozakira Zulfa (Universitas Malikussaleh)
Fadli Mahyudin (Universitas Malikussaleh)
Arnawan Hasibuan (Universitas Malikussaleh)

Indonesian Journal of Applied Informatics
Vol 10, No 1 (2025)

Abstrak : Stunting merupakan masalah gizi kronis yang berdampak signifikan terhadap pertumbuhan dan perkembangan anak, khususnya pada periode 1000 hari pertama kehidupan. Penelitian ini mengembangkan sistem berbasis web untuk memprediksi risiko stunting pada anak menggunakan metode Logistic Regression. Dataset mencakup fitur seperti jenis kelamin, usia, berat dan panjang lahir, berat badan, panjang badan, serta status konsumsi ASI. Data diproses melalui tahapan cleaning dan preprocessing sebelum pelatihan model. Logistic Regression dipilih karena kemampuannya melakukan klasifikasi biner, yaitu stunting atau tidak. Penilaian model dilakukan dengan memanfaatkan metrik accuracy, precision, recall, dan AUC. Hasil eksperimen menunjukkan model mencapai accuracy 69,9%, precision 78,4%, recall 85,4%, dan AUC 0,94, sehingga dapat digunakan sebagai alat bantu untuk deteksi dini stunting. Sistem berbasis web ini memudahkan akses bagi tenaga kesehatan dan orang tua untuk memantau status gizi anak dan memberikan rekomendasi intervensi yang tepat. Temuan ini menunjukkan bahwa penerapan Logistic Regression efektif dalam mendukung pencegahan stunting secara lebih terarah.================================================Abstract : Stunting is a chronic nutritional problem that significantly affects children's growth and development, particularly during the first 1,000 days of life. This study develops a web-based system to predict the risk of stunting in children using the Logistic Regression method. The dataset includes features such as gender, age, birth weight and length, body weight and length, and breastfeeding status. The data were processed through cleaning and preprocessing stages before model training. Logistic Regression was chosen for its ability to perform binary classification, i.e., stunted or not stunted. The model was evaluated using accuracy, precision, recall, and AUC metrics. Experimental results show that the model achieved accuracy of 69.9%, precision of 78.4%, recall of 85.4%, and AUC of 0.94, indicating its potential as a tool for early detection of stunting. This web-based system facilitates access for healthcare workers and parents to monitor children's nutritional status and provide appropriate intervention recommendations. The findings demonstrate that Logistic Regression is effective in supporting more targeted stunting prevention efforts.