Exploratory Data Analysis To Identify The Most Important Feature Of University Admission Test Criteria Using Random Forest And Neural Network Algorithm

The University institutions currently face challenges in conducting the admissions process to get new students. The process must be carried out in the right way to ensure that prospective students who are accepted have the right abilities to meet academic targets in their chosen scientific field. The admission process is carried out based on predetermined criteria by determining the weight of the requirements according to the policy for that period. There is often a mismatch in the abilities of the student candidates who are accepted with the skills needed in the chosen field, so that there is a potential dropout risk for these students. One way to avoid this is to know the essential criteria in the admission test. We create two models based on the chosen algorithm. The Random Forest algorithm has a better accuracy rate, which is 85.17%, compared to the 80,27% accuracy rate of the Neural Network algorithm. This study found the most important feature of the admission process is the school ranking, where this feature has the most significant influence compared to the other, which is more than 20% importance's rate. This study also found a significant difference in the gender distribution of the accepted applicants, with a ratio of 24% for male registrants and 76% for female registrants. For the failed registrants, there were 25% male registrants and 75% female registrants. This study is based on the admission test data so that the most important feature found in this study can be used as a basis for policymaking for admission tests to come.