Abstrak - Stunting adalah malnutrisi yang ditandai dengan tinggi badan, diukur dengan standar deviasi dari WHO. Dinas Kesehatan Provinsi Gorontalo khususnya dibidang Gizi mengenai stunting, selama ini melakukan kegiatan pemantauan tiap-tiap puskesmas dan posyandu. Pemantauan dan pendataan terkait stunting di berbagai puskesmas faktor penting. Masalah yang sering muncul adalah data yang dikumpulkan untuk underestimasi selalu tidak akurat setiap bulannya, karena hanya perkiraan yang dihitung berdasarkan kasus puskesmas. Prediksi yang akurat diperlukan untuk mengatasi permasalahan yang ada. Penelitian ini menggunakan algoritma K-Nearest Neighbor (K-NN) menggunkan Chi Square. Berdasarkan hasil eksperimen, prediksi jumlah penderita stunting telah berhasil dilakukan. Maka nilai hasil dari prediksi tersebut dapat diimplementasikan untuk bahan pertimbangan atau kebijakan didalam pengambilan keputusan. Tingkat error terkecil hasil RMSE 1,200 pada algoritma K-Nearest Neighbor menggunkan Chi Square dibandingkan algoritma K-Nesrest Neighbor tanpa seleksi fitur. Dengan demikian dari hasil eksperimen menunjukan bahwa penambahan seleksi fitur telah menunjukan performa kinerja yang baik pada algoritma K-Nearest Neighbor.Kata kunci: Prediksi, Stunting, K-NN, Chi Square Abstract - Stunting is a nutritional deficiency characterized by height as measured by the WHO standard deviation. The Gorontalo Provincial Health Office, especially in the field of nutrition related to stunting, has so far carried out monitoring activities at every puskesmas and posyandu. Monitoring and data collection related to stunting in various health centers is an important factor. The problem that often arises is that the data collected for underestimation is always inaccurate every month, because only estimates are calculated based on puskesmas cases. Accurate predictions are needed to overcome the existing problems. This study uses the K-Nearest Neighbor (K-NN) algorithm using Chi Square. Based on the experimental results, the prediction of the number of stunting sufferers has been successfully carried out. Then the value of the predicted results can be implemented for consideration or policy in decision making. The smallest error rate is RMSE 1,200 in the K-Nearest Neighbor algorithm using Chi Square compared to the K-Nesrest Neighbor algorithm without feature selection. Thus, the experimental results show that the addition of feature selection has shown good performance on the K-Nearest Neighbor algorithm.Keyword: Prediction, Stunting, K-NN, Chi Square
Copyrights © 2022