Multiclass Prediction Model for Student Grade Prediction Using Machine Learning

Abstract

Today, predictive analytics applications became an urgent desire in higher educational institutions. Predictive analytics used advanced analytics that encompasses machine learning implementation to derive high-quality performance and meaningful information for all education levels. Mostly know that student grade is one of the key performance indicators that can help educators monitor their academic performance. During the past decade, researchers have proposed many variants of machine learning techniques in student grade prediction. However, there is a lack of studies in identifying the effective predictive model, especially in addressing imbalanced multi-classification for student grade prediction. Therefore, this paper presents a comprehensive analysis of machine learning techniques to predict the final student grades in the first semester courses by providing better prediction accuracy. Two modules will be highlighted in this paper. First, we compare the accuracy performance of five well-known machine learning techniques, namely Decision Tree (J48), Support Vector Machine (SVM), Naïve Bayes (N.B.), K-Nearest Neighbor (kNN), and Logistic Regression (L.R.), to our real dataset. Second, we proposed a multi-class prediction model for an imbalanced multi-class dataset using two types of data-level solutions to improve the prediction accuracy performance. The obtained results indicate that the proposed Synthetic Minority Oversampling Technique (SMOTE) and wrapper-based feature selection (F.S.) integrates with kNN shows significant improvement with the highest accuracy of 99.6%. In contrast, all F.S. algorithms perform 99.4% robust performance when working with SVM independently. These findings indicate that the proposed model generally discovers various F.S. algorithms and SMOTE in addressing imbalanced multi-classification, which will help the researcher to improve the performance for student grade prediction.

Prediction of College Students’ Sports Performance Based on Improved BP Neural Network

“…Backpropagation (BP) neural network is a highly complex and nonlinear dynamic analysis system proposed by Rumelhart and McClelland in 1986. It forms an interconnected network structure through various independent units [11][12][13][14][15]. e neural function learns the data samples, establishes the connection weights and thresholds between the units, and then deals with complex nonlinear problems without specific functional forms [16][17][18].…”

Is Initial Performance in a Course Informative? Machine Learning Algorithms as Aids for the Early Detection of At-Risk Students

“…Our research suggests that particular machine learning algorithms can be used to make informed predictions regarding students' performance attainment, but the predictive validity of each algorithm has to be first assessed as a function of two important variables: course type and instructional mode. Our study adds to the growing body of grade prediction studies that rely on machine learning algorithms [8,10,[43][44][45] by pointing to the relevance of such variables to interventions that are intended to foster academic success in an understudied student population. In our research, the latter is represented by young women of college age from a society that has only recently implemented and enforced gender equity guidelines.…”

“…To wit, unrecognized difficulties (an event classified in signal detection theory as a miss) are likely to lead to course failure. Notwithstanding the need for valid early predictions of students' academic performance, which rely on limited information, most of the research on algorithms that are intended to assist educators' performance forecasts has relied upon much greater amounts of information collected within a much larger timeframe and has often involved discipline-specific subject matters [7][8][9]. Examples are predictions of final course grades in a particular subject matter based on students' grade point average (GPA), as well as grades in pre-requisite courses [10], or more simply, on students' academic history, as exemplified by their performance in past courses [11].…”

Predictive modelling and analytics of students’ grades using machine learning algorithms

“…Though unbalanced, an accuracy as high as 99% was obtained for identifying a student as pass, fail or drop out. (Bujang et al, 2021 ) and (Costa et al, 2017 ) catered for imbalance classes during their analysis. (Bujang et al, 2021 ) indicated how the combination of SMOTE and feature selection influence accuracy of predictive models.…”

“…(Bujang et al, 2021 ) and (Costa et al, 2017 ) catered for imbalance classes during their analysis. (Bujang et al, 2021 ) indicated how the combination of SMOTE and feature selection influence accuracy of predictive models. (Costa et al, 2017 ) coupled SMOTE with fine tuning of algorithms.…”