Optimization Of Support Vector Machine (Svm) Based Forward Selection For Prediction Of Incoming Students Continue To Private College

Abstract

Abstract.

     The large volume of society can cause problems if it is not commensurate with improving the quality of human resources. A factor that can support human resource capacity is improving the quality of education. High school student data has quite diverse data. With a case study at a high school in Brebes Regency, this experiment is used as a basis for predicting the distribution of high school graduates in the following year. The data mining process is assisted by the WEKA application. The classification used is a support vector machine classification based on forward selection to determine the attributes that are most influential in prediction. The highest results from the SVM experiment were obtained by kernel anova with an accuracy value of 96.17%. Then the FS-SVM algorithm with anova kernel parameter C of 0.5 with an accuracy level of 99.71%.

Keywords: High School, Course, Data, Classification