<![CDATA[Competition in the higher education, especially private higher education (PTS) in the digital era, is becoming increasingly tough. In order to achieve the number of prospective new students, various methods are used so that the target for admitting the number of new students can be achieved in each new academic year. Providing a scholarship program is one way to attract the prospective new students. The awarding of a scholarship program must consider various possibilities such as the seriousness or commitment of the prospective new student. Refusal to grant scholarship programs can occur and become an obstacle for achieving the target. The prediction model through machine learning using some variables such as high school’s name, high school “category”, province or area of high school located, focus of specialization in high school, high school’s grade, type of parents income, and selected major of study in higher education. All of those variables will provides the probability values that will become an indicator that can be used to prioritize requests for scholarship program applications by taking into account the factors of acceptance or rejection from prospective students. Currently there is no measurement with accuracy of acceptance or rejection from prospective students. The purpose of this research is to build and compare machine learning models such as Logistic Regression, Artificial Neural Networks, Support Vector Machines, Decision Trees, Naïve Bayes, and K Nearest Neighbors so that a machine learning model is obtained that has the best predictions for awarding scholarship programs. The result of this research is that the Logistic Regression model has the highest model average accuracy value (62,05%) from training data compared to others. The highest accuracy of Logistic Regression model (62,29%) achieved based on the testing data. The highest AUC value (0,818) generated by Logistic Regression model which means the model is able to do the classification categorized “Good Classification” compare to other models.]]>
