Analysis classification of households who received "raskin" in Semarang City using Fuzzy K-Nearest Neighbor (FKNN) and Support Vector Machine (SVM)

Abstract

Data mining or Big Data is a very important part of going to the industrial revolution era 4. Data mining is inseparable from statistical analysis for classification methods. Data mining is data with a very large size. Two of the methods in data mining is classification using Fuzzy K Nearest Neighbor (FKNN) and Support Vector Machine (SVM). The concept of FKNN is based on fuzzy members, while the SVM method is based on a hyperplane. In this study, the classification of poor rice receipt in the city of Semarang used the FKNN and SVM methods. These methods applied to classification the household wom receipt “raskin” in Semarang city, Indonesia. “raskin” is one of Indonesia government program to assist the households who categorized in the poor households. We used some variables independent such as the characteristic of house and the criteria of head households. The data collected from SUSENAS-social economic survey 2016 in Semarang city with 930 households. From the results of the analysis, it was found that the characteristics of residential houses more influenced the factors of “raskin” revenue compared to the characteristics of the head of the household. The SVM method produces better accuracy than FKNN. The best accuracy value reaches 90% with the radial base function kernel function.