Towards a Framework for Performance Management and Machine Learning in a Higher Education Institution

Abstract

This paper proposes a new structure for management and machine learning in higher education institutions, which is designed to improve the efficiency of an organization and the success of the students at a whole. The framework brings about the enactment of several analytical techniques, like predictive modeling and data-driven decision making, which help to make accurate strategies for planning and providing continuous improvement. Four algorithms in machine learning- Linear Regression, Decision Tree, Random Forest and Multilayer Perceptron- are compared to see if they predict important performance markers for student success, faculty productivity and institutional efficiency. The results illustrate the Multilayer Perceptron algorithm as the best performer, getting MSE of 0.018 and MAE of 0.105, while R2 score of 0.842, showing the superiority of MLP over the others. Validation studies done comparing it with base line models or related models in the field are proof that the suggested model is widely applicable among the higher education spectrum in dealing with the involved issues. The imaginable framework seems to be a prospective tool for stimulating creativity, inclusion, and eminence in academia while adding to the knowledge acquisition and achieving institute objectives.