Pengujian akurasi model regresi logistik multinomial untuk memprediksi keberhasilan mahasiswa di perguruan tinggi menggunakan r

Abstract

Higher education institute should have maintained its student’s success in their academic field. Therefore, higher education institute should make a model to predict its success as early as possible. To have such model, the impacting factors should be determined, which factor are in the forms of continuous dan categorized data. This research is aimed to build logistic regression model based on compounded data from continuous and categorized data, which then test the model accuracy to predict the student’s success in such institute. Research data is using 68 student’s data. There are 6 research steps, first, data preparation and collection, second, data analysis, third, building the logistic multinomial regression model, fourth, data testing and validation, fifth, measure the model accuracy, and the sixth step is drawing the conclusion based on the analysis output. The result of prediction analysis dan accuracy test using logistic multinomial regression is the best model with significant factors that influence the study time are gender, department and selection track, while prediction accuracy model for each study time response variable is 96.4%.