Evaluation of Students’ Performance in Mathematics using Machine Learning Algorithms in Federal College of Education (Special), Oyo, Nigeria

Abstract

This paper focuses on evaluation of students' performance in Mathematics using Machine Learning Algorithms in Federal College of Education (Special), Oyo, Nigeria. The study specific objectives are to: determine how machine learning algorithms enhance the evaluation of students' performance in Mathematics, determine how classifier metrics enhance the evaluation of students’ performance in Mathematics. Two research hypotheses guided the study. Waikato Environment for Knowledge Analysis (WEKA) tool was used to evaluate students’ performance in Mathematics which contains various machine learning algorithms. The population for the study comprised 6,982 students from 100 level, 200 level, and 300 level respectively. The sample size for the study consists of 3,642 students across the level. The instrument used for data collection was a Google form questionnaire, The following algorithms J48, Logistic Model Tree, Random Forest, Random Tree, RepTree, and Hoeffding Tree algorithms were used to evaluate students' performance in Mathematic and the following evaluation metrics were considered: Recall, Precision, F-measure, True-positive rate and False-positive rate. The results of the classification model indicated that the Random tree algorithm provides a better classification accuracy of 99.014%. The conclusion was arrived that the Random tree algorithm is the most suitable algorithm for performance evaluation because it yielded the best result in terms of accuracy, precision, recall, True False, False True, F-measure and other evaluation metrics. It was recommended among others that teachers should be adequately equipped by organizing workshops which will enable them to create an enabling environment for learners to be able to learn Mathematics effectively.