PREDICTIVE ANALYTICS FOR STUDENTS’ ACADEMIC PERFORMANCE USING MACHINE LEARNING (CASE OF STUDY AGP T/M STUDENTS)

Abstract The aim of every educational setup is giving the best educational experience and knowledge to the students. Discovering the students who need extra support and guidance so as to carry out the necessary actions to enhance their performance plays an important role in achieving that aim. Researchers have proposed various machine learning methods to predict students’ performance but still improvement is needed because some are not giving best results due to non-effective cleaning of data in the preprocessing stage and the feature selected. In this research work, five machine learning algorithms have been used to predict the performance of the students in higher institutions considering one Institution which is Abdu Gusau Polytechnic, Zamfara State. And the limitations from existing researches were treated. The machine learning algorithms include Random Forest, Support Vector Machine, Logistic Regression, K-Nearest Neighbour and Naive Bayes algorithms and thorough cleaning and effective feature selection was made to improve the performance of the algorithms. The classifiers have been compared and evaluated using Precision, Recall, F1-Score and Support classification accuracy. The dataset used to build the models were collected from the MIS departments of each of the institution. The result indicated that Random forest classifier achieved the best performance that is equal to 76.56% while Support Vector Machine, Logistic Regression, K-Nearest Neighbour and Naïve Bayes gave 73.21%, 75.60%, 71.29%, 33.49% respectively. Keywords: Random Forest, Support Vector Machine, Logistic Regression, K-Nearest Neighbour and Naive Bayes, Students' Performance Prediction, Feature selection