PERBANDINGAN METODE KLASIFIKASI NAÏVE BAYES DAN K-NEAREST NEIGHBOR PADA ANALISIS DATA STATUS KERJA DI KABUPATEN DEMAK TAHUN 2012

Abstract

Large population in Indonesia is closely related to the working status of the population which is unemployed or employed. It can lead to the high unemployment when the avaliable jobs arent balance with the population. Used two methods to perform the classification of employment status on the number of residents in the labor force in Demak for 2012 which is Naïve Bayes and K-Nearest Neighbor. Naïve Bayes is a classification method based on a simple probability calculation, while the K-Nearest Neighbor is a classification method based on the calculation of proximity. Variables used in determining whether a person's employment status is idle or not are gender, status in the household, marital status, education, and age. Employment status of the data processing methods of Naïve Bayes with the accuracy obtained is equal to 94.09% and the K-Nearest Neighbor method obtained is equal to 96.06% accuracy. To evaluate the results of the classification used calculations Press's Q and APER. Based on the analysis, the Press's Q values obtained indicate that both methods are already well in the classification of employment status data in Demak. Based on the calculation of APER, the classification of data in the employment status of Demak using the K-Nearest Neighbor method has an error rate smaller than the Naïve Bayes method. From this analysis it can be concluded that the K-Nearest Neighbor method works better compared with the Naïve Bayes for employment status data in the case of Demak for 2012. Keywords: Classification, Naïve Bayes, K-Nearest Neighbor (K-NN), Classification evaluation