The Role of Machine Learning in Identifying Students At-Risk and Minimizing Failure

Abstract

Education is very important for students' future success. The performance of students can be supported by the extra assignments and projects given by the instructors for students with low performance. However, a major problem is that students at-risk cannot be identified early. This situation is being investigated by various researchers using Machine Learning techniques. Machine learning is used in a variety of areas and has also begun to be used to identify students at-risk early and to provide support by instructors. This research paper discusses the performance results found using Machine learning algorithms to identify at-risk students and minimize student failure. The main purpose of this project is to create a hybrid model using the ensemble stacking method and to predict at-risk students using this model. We used machine learning algorithms such as Naïve Bayes, Random Forest, Decision Tree, K-Nearest Neighbors, Support Vector Machine, AdaBoost Classifier and Logistic Regression in this project. The performance of each machine learning algorithm presented in the project was measured with various metrics. Thus, the hybrid model by combining algorithms that give the best prediction results is presented in this study. The data set containing the demographic and academic information of the students was used to train and test the model. In addition, a web application developed for the effective use of the hybrid model and for obtaining prediction results is presented in the report. In the proposed method, it has been realized that stratified k-fold cross validation and hyperparameter optimization techniques increased the performance of the models. The hybrid ensemble model was tested with a combination of two different datasets to understand the importance of the data features. In first combination, the accuracy of the hybrid model was obtained as 94.8% by using both demographic and academic data. In the second combination, when only academic data was used, the accuracy of the hybrid model increased to 98.4%. This study focuses on predicting the performance of at-risk students early. Thus, teachers will be able to provide extra assistance to students with low performance.

Closing the gap: exploring the untapped potential of machine learning in deaf students and hearing students’ academic performance

“…Hybrid deep learning models enhance the modelling process in terms of computation, functionality, resilience, and accuracy. From the major works listed in Table 2, the significant performance is resulted in [34,40,45,50,53]. These studies show that the performance accuracy of the models can be very high, though it depends on the dataset quality.…”

“…Prediction of at-risk students [31,32,34,65,38,40,41,42,52,67,63,66,62,60,90,91,61,58,59,64] 20 Features affecting the academic performance [76,36,39,43,44,79,70,77,51,81,80,69,83,74,82,78,84,92,68,85,75,71,93,72] 24 Suitable Course selection [87,86,55,88,89,48] 6 Prediction of marks / grades [29,30,37,45,46,47,94,49,50,53,95,96,62,90,97,98,58,99,100,16,72,9 8,18,15,75,101,102,64]<...…”

Predicting Student Performance in Online Learning: A Multidimensional Time-Series Data Analysis Approach

“…This method utilized a genetic algorithm to capture the 30 best attributes from students' historical learning data and trained a K-NN regression model and a decision tree using these features and labels to predict students' performance score and categories. PEK et al [16] used naive Bayes, random forest, decision tree, AdaBoost classifier, logistic regression, and KNN algorithms as basic learners, support vector machine (SVM) as meta-learner, and created a stacking method to develop a hybrid ensemble model. By analyzing the data, they discovered important features that influence student learning outcomes and successfully helped teachers identify students at risk.…”

Exploring the role of machine learning in forecasting student performance in education: An in-depth review of literature

“…The number of records assessed for eligibility was 709. After screening, highly relevant studies were identified (n = 50) (Adnan et al, 2021(Adnan et al, , 2022Alamri & Alharbi, 2021;Alhazmi & Sheneamer, 2023;Aljaloud et al, 2022;Alshanqiti & Namoun, 2020;Asthana et al, 2023;Bujang et al, 2021;Chan et al, 2023;Chui et al, 2020;Feng et al, 2022;Gao et al, 2019;Ghorbani & Ghousi, 2020;Hassan et al, 2022;Hussain et al, 2021;Kastrati et al, 2020;Kusumawardani & Alfarozi, 2023;Latif et al, 2023;Liu et al, 2020;Mengash, 2020;Motz et al, 2021;Munshi & Alhindi, 2021;Nabil et al, 2021;Nguyen-Huy et al, 2022;Okoye et al, 2023;Orlando et al, 2020;Pek et al, 2023;Prabowo et al, 2021;Rafique et al, 2021;Rahman et al, 2021.…”