Perbandingan Metode Klasifikasi Naive Bayes dan K-Nearest Neighbor

Abstract

Classification is a technique to build a model and assess an object to put in a particular class. Naive Bayes is one of algorithm in the classification based on the Bayesian theorem, which assumes the independencies of one class with another class. K-nearest neighbor is an algorithm in the classification method for classifiying based on data that has a closest distance between one object and another object. Naive Bayes and k-nearest neighbor methods are used in classification of the employment status of citizen in Kutai Kartanegara regency because has a good accuracy and produce a small error rate when using large data sets. This research aim to compared optimal performance accuracy of both methods on the classifiying of the employment status of citizen. The data used are employment status of citizen in Kutai Kartanegara Regency based on SAKERNAS of East Kalimantan Province in 2018 and used 5 factors namely age, sex, status in the household, marital status, and education to predict employment status of citizen. Based on the analysis, classification the employment status of citizen with naive Bayes method has accuracy of 90,08% and in the k-nearest neighbor has accuracy of 94,66%. To evaluate the accuracy of classification used calculation of Press’s Q. Based on Press’s Q value showed that both of classification methods are accurate. From that analysis, can be concluded that the k-nearest neighbor method works better compared with the naive Bayes method for the case of the employment status of citizen in Kutai Kartanegara Regency.