Analysis and Prediction of Graduate Admissions Based on Pre-COVID and post-COVID Scenario

Abstract

Graduate admissions is one of the events that attracts a lot of attraction from prospective students and universities alike. Be it the university conducting graduate admissions or an aspiring student; both yearn for a prediction system to aid in the process of selecting admits. On one hand, the university can get an insight on the probability of a student's admit thus aiding the graduate admissions office in their workload, and on the other hand the student can get a forecast on the chance of admit and can take preemptive decisions to facilitate the process. However, due to the COVID-19 pandemic, the graduate admissions has seen a slight change in paradigm. This change creates confusion among the related masses. A probing analysis on this change serves as a reference to act upon. In this study, prediction models are built with an extra parameter signifying whether a record in the dataset belongs to the COVID-19 pandemic period. Various models such as Logistic Regression, Decision Tree, Random Forest, Gaussian Naive Bayes and Artificial Neural Networks are used to determine the change in probability of admission due to the effect of the pandemic. All the models provide an accuracy score in the range of about 55% to 80%, with the Neural Network outperforming all the other models with a test accuracy score of 79.03%. The effect of the pandemic has caused an ambiguous response to various factors, but it can be stated the chances of admits of students have generally increased likely due to the lower number of applicants