The definitive goal of any educational institute is to provide students with the best possible educational knowledge and practice. Categorizing students who need additional backing and taking proper steps to improve student performance plays a key role in attaining this goal. A student's academic performance is a significant aspect for educational success at all stages. Academic background is very important for students who want to continue their studies and secure their future. Several studies have uncovered factors related with personal responses such as: For example, in relation to family’s communication, understanding and anticipating student perspectives on campus to improve student performance. A regressor that can forecast student success in the academic’s status from the student dataset collected from online repositories. The main goal of this research work focus on the academic performance of the students through their behavior using five different Machine Learning techniques such as Random Forest Regressor (RFR), Extra Trees Regressor (ETR), Gradient Boosting Regressor (GBR), Bagging and ElasticNet. The comparative analysis of these models shows that RFR model best fit model through the predictors lazy regressor and standard scalar methods. A proposed model is developed by hyperparameter tuning the best fit model using random search. The performance metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), R Squared, Adjusted R Squared shows that proposed method is better than the existing models.

Hyperparameter tuning, Machine Learning, Random Forest, student performance, lazy regressor, random search