Klasifikasi Pengaruh Polusi Udara di Indonesia terhadap Kesehatan menggunakan Algoritme Kernel Modified K-Nearest Neighbor

Fayza Sakina Maghfira Darmawan (Fakultas Ilmu Komputer, Universitas Brawijaya)
Imam Cholissodin (Fakultas Ilmu Komputer, Universitas Brawijaya)
Putra Pandu Adikara (Fakultas Ilmu Komputer, Universitas Brawijaya)

Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Vol 6 No 6 (2022): Juni 2022

Air pollution has many problems, one of which is health problems. The effect of air pollution on health can have a mild, severe, and even death effect. The solution to this problem is to classify so that the amount of levels that cause air pollution gives the appropriate results of influence. The method used in this study is the Kernel Modified K-Nearest Neighbor (KMKNN) algorithm. KMKNN is a modified algorithm of KNN (MKNN) that uses kernel distance. The data used is 480 data with three features and twelve classes. In the research test, the data sharing methods used were K-Fold Cross Validation and Hold-Out. The test was performed using three different kernels. The best constant of the RBF kernel is 0,97 at constants 1 through 5. The best parameter degree of polynomial kernels is 0,97 at degrees 1 and 2. The results of the K-Fold test with each kernel provide an average accuracy at 0,87. The Hold-Out test results of each kernel provide the highest average accuracy of 70%:30% split at 0,972.