PREDICTING THE GRADES OF STUDENTS BASED ON THEIR BEHAVIOURS USING MACHINE LEARNING

Abstract

A serious menace to the community is alcoholism. Nowadays, there is a significant problem with student drinking. Poor academic performance in students is a result of alcohol addiction. It has been suggested that common risk factors such as unstable homes, poor mental health, and unsupportive families may make teenagers more likely to use alcohol and do poorly in school. Excessive drinking among college students is linked to several detrimental outcomes, such as teenage suicide, fatal and nonfatal injuries, violence, academic failure, sexually transmitted infections, rape and assault, and unwanted pregnancy. A comparison experiment on predicting alcohol intake among college students is described in the study.
In this study, we analysed a dataset made up of student characteristics, grades, and alcohol consumption levels. For this research, we are utilising machine learning methods. This study seeks to determine the predictive value and relevance of the variable. We use the logistic regression approach to extract the variable significance from the data, and depending on the other characteristics, we apply models to the dataset using the following algorithms: Linear Regression, Decision Tree Regressor, Extra Trees Regressor, Random Forest Regressor, and SVM. We discovered that Random Forest Regressor has the highest R2 score of all the models i.e., 91.2945, thus we went on to use this model to predict the final grades.