PERBANDINGAN METODE PENGUKURAN JARAK PADA ALGORITMA K-NEAREST NEIGHBOR UNTUK KLASIFIKASI
Daftar Isi:
- K-Nearest Neighbor is a non-parametric classification algorithm that does not use training data and does not use initial assumptions or models in the calculation process. The quality of the classification results of the k-Nearest Neighbor algorithm is very dependent on the distance between object and value of k specified, so the selection of method for distance measurement determines the results of classification. In this study, several methods of measuring Euclidean distance distance, Manhattan distance, Tchebychev distance and Cosine distance were examined to see distance measurement methods that can be used optimally on the k-Nearest Neighbor algorithm. The selection of k values also determines the results of the classification of the k-Nearest Neighbor algorithm, for which the determination of the value of k also needs to be considered. This study uses a cervical cancer dataset and provides the highest accuracy results in the Cosine distance distance measurement method that is equal to 92.559% at the value of k=9, while for the Manhattan distance measurement method is 91.666% with a value of k=13, the Tchebychev distance measurement method is 91.666% with the value of k=13 and the lowest accuracy value obtained by the Euclidean method that is equal to 91.071% at the value of k=13. Based on the value of accuracy, the most compatible distance measurement method used is Cosine distance even with the best k value obtained is k=9 though it has the highest computation time which is 0.898 s compared to the Euclidean method of 0.555 s, the Manhattan method is 0.5882 s and Tchebychev has the lowest computation time which is 0.49 s.