A Stress Based Prediction Model for University Student Using Support Vector Machine and Grid-Search-CV for Parameter Turning

Abstract

The current academic system consists of various mental struggles ranging from family, peers, lecturers and the academic system generally. However, high level of stress on university students negatively affect their academic performance. In this paper, we describe how to efficiently select the best parameters to develop the proposed model. The Grid-Search-CV techniques is adopted to fine-tune the Support Vector Machine(SVM) classifier with different parametric combination, the best parameter configuration that provides the highest prediction accuracy is selected for training our model. Hence, the proposed student stress prediction model has shown a high degree of prediction accuracy (99%).