A Robust Prediction Model for Candidate’s Admission using Fletcher-Reeves (FR) Conjugate Gradient Method
Selection of prospective candidates to a higher institution of learning based on candidate’s choice of course of study is now becoming a herculean task, particularly when the carrying-capacity of two hundred and fifty–six higher
institutions in Nigeria cannot admit over one million eligible candidates seeking admission yearly. Several works
have been done in the past using Nave Bayes algorithm, Decision trees, K-Means algorithm, Random Forest and
other Machine learning algorithms to predict Candidate’s admission to higher institutions. Previous methods were
confronted with pockets of shortcomings. These include required lengthy offline/batch data training, unable to learn incrementally or interactively in real-time, poor transfer of learning ability, and non-reusability or integration of modules etc. In this paper, a machine learning model was implemented using Fletcher-Reeves Conjugate Gradient algorithm to predict candidates’ selection into a higher institution of their choice. The algorithm was implemented using python programming language. The algorithm was found to perform better than the Gradient Method with 89% prediction accuracy compared to 83% prediction accuracy of the gradient method.
Keywords:Candidate’s Admission, Neural Networks, Optimization Algorithm, Prediction Model