The ensemble approach such as boosting is based on heuristic system to develop prediction paradigms. The ensemble learning techniques are typically more accurate than individual classifiers to generate predictions. For these grounds, primarily in this study several ensemble techniques have been discussed to get a comprehensive knowledge of key methods. Among various ensemble approaches, researchers have practiced boosting mechanism to predict the performance of students. As application of ensemble methods is contemplated to be significant phenomenon in classification and prediction procedures, therefore the researchers exploited boosting technique to develop an accurate prediction pedagogical model, in view of the pronounced nature and novelty of the proposed method in educational data mining. The base classifiers including random tree, j48, knn and naïve bayes have been evaluated on 10- fold cross validation system. Moreover, filtering procedures such as oversampling (SMOTE) and under-sampling (Spread subsampling), have been exploited to further inspect any significant change in results among meta and base classifiers. Both ensemble and filtering approaches have demonstrated substantial improvement in predicting the performance of students than the application of conventional classifiers. Furthermore, based on the improvement in results two novel prediction models have been propounded after conducting performance analysis on each approach.