Perbandingan Akurasi Antara Metode K-Nearest Neighbor (KNN) Dan Artificial Neural Network (ANN) untuk Klasifikasi Indeks Pembangunan Manusia Kabupaten/Kota di Pulau Jawa

In 2021, 56.1% of Indonesia's population was on the Java Island, so the government needs to do mapping for policy making in various fields. Classifying the Human Development Index (HDI) can be done to assist the government in measuring the results of human resource development. The purpose of this research is to compare the accuracy results of two methods, namely K-Nearest Neighbor (KNN) and Artificial Neural Network (ANN) to classify the HDI of districts/cities in Java Island. The results showed that the application of KNN and ANN methods on the same data resulted in different accuracy values. In the KNN method, using 80%-20% training and testing data, the K=7 value shows the highest accuracy rate [95.83%]. While the ANN method with the split 70%-30%, resulting in the highest accuracy value [94.44%]. By calculation, KNN method produces a higher accuracy value. However, the evaluation results using Fold Cross Validation show that the best KNN model is at K=3, with a mean score 84.85%. While in the model of the highest accuracy value of the ANN method, there is overfitting. Based on this comparison, it can be concluded that the highest accuracy value of both KNN and ANN methods both have weaknesses.