Improving Prediction Capability of Ensemble of Classifiers through Weighted Average Probabilities
Princy Christy. A1, N. Rama2

1Princy Christy. A, Research Scholar, Post Graduate and Research Department of Computer Science, Presidency College, Chennai (Tamil Nadu), India.
2N. Rama, Principal, Dr. M.G.R. Government Arts and Science College for Women, Villupuram (Tamil Nadu), India.

Manuscript received on 18 June 2019 | Revised Manuscript received on 25 June 2019 | Manuscript published on 30 June 2019 | PP: 2540-2543 | Volume-8 Issue-5, June 2019 | Retrieval Number: E7829068519/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Ensemble of classifiers has been proved to significantly impact the performance outcome by improving the capability of performance classification in education scenario. An approach, Optimised Voting Ensemble (OVE) has been proposed to increase the prediction accuracy in determining the final class of students based on their continuous performance. The base learners chosen for this approach is decision tree, multi layer perceptron and stocastic gradient descent classifiers that are well suited for a multi-class problem. The approach tends to improve the classification and prediction accuracy by optimization of hyperparameters. The hyperparameters of base learners are tuned such that they improve their capability to classify and predict the final grade. The optimization is carried out through grid search. A weighted average probability method is used to combine the tuned base learners to form an ensemble. This tends to combine the strength of all the base learners and improve the prediction accuracy by assigning weight to strong classifiers.
Keywords: Optimization, Student Performance, Voting Ensemble, Hyperparameter, Prediction

Scope of the Article: Discrete Optimization