Machine Learning Approach for Scholarship Candidate Selection in Indonesia Mohamad Irfan (a), Wildan Budiawan Zulfikar (a*), Rina Anjari Ramadanti (a), Agung Wahana (a), Yana Aditia Gerhana (a)
a) Department of Informatic, UIN Sunan Gunung Djati, Bandung 40614, Indonesia
*wildan.b[at]uinsgd.ac.id
Abstract
Scholarships are a form of assistance in the form of funds provided by an institution to individuals, both students and students, with the aim of being used for the continuation of the education they are pursuing. Administratively, candidates register and submit the required files. The selection process must be carried out objectively. The selected participants must truly meet various requirements and have more rights than other participants. The problem that arises is the issue of subjectivity in this selection process. In this research, two machine learning algorithms are implemented and compared. The two algorithms are Modified K-Nearest Neighbor (MKN) and Classification and Regression Tree (CART). MKNN is an improvisation of the K-Nearest Neighbor method where the calculation process includes data validity steps and a weight voting process. Based on related studies, CART is a classification algorithm that is easy to interpret but has good accuracy. Based on the limited evaluation, MKNN recorded an accuracy value of 91.23%, while for the CART method it was 87.7%.
