Student’s Performance Analysis and Prediction with the Help of Machine Learning Models

Abstract

Educational Data Mining (EDM) has become one of the most important fields now a day because, with the development of technology, student’s problems are also increasing. These problems can be related to student’s behavior, parents' participation or academic etc. To tackle these problems and help students, educational data mining has come into existence. The prior prediction of student’s performance is necessary so that useful steps can be taken out to help him and guide him in the correct direction. In this research, a systematic literature review (SLR) has been performed to get 20 studies (2012-2019) in the area of EDM. The reason behind SLR is to get advanced machine learning models that have been used by researchers in their field so that we can compare them with each other to get the most useful model among them. Feature extraction and data augmentation techniques will be used to enhance their performance and predictions. After detailed SLR, 11 highly advanced machine learning models have been obtained. These models are further applied to 2 public databases to check their performance and prediction rate. It is observed that “Random forest” and “Decision tree” are the best machine learning models having an accuracy score of 95% and 96% respectively. To validate results, database had been splited into 70/30 ratio. 70% database was used to train models and remaining 30% database was used to test and validate results. With the help of these experiments, weak students can be easily identified and proper precautions can be taken to help them. In future work, these models can be implemented on real-time university or school databases to further enhance their accuracy and performance scores. With the help of student performance, student’s areas of interest and the future job can also be predicted.