Prediction of Student Academic Performance Using a Hybrid 2D CNN Model

Abstract

Opportunities to apply data mining techniques to analyze educational data and improve learning are increasing. A multitude of data are being produced by institutional technology, e-learning resources, and online and virtual courses. These data could be used by educators to analyze and understand the learning behaviors of students. The obtained data are raw data that must be analyzed, requiring educational data mining to predict useful information about students, such as academic performance, among other things. Many researchers have used traditional machine learning to predict the academic performance of students, and very little research has been conducted on the architecture of convolutional neural networks (CNNs) in the context of the pedagogical domain. We built a hybrid 2D CNN model by combining two different 2D CNN models to predict academic performance. Our sample comprised 1D data, so we transformed it to 2D image data to test the performance of our hybrid model. We compared the performance of our model with that of different traditional baseline models. Our model outperformed baseline models, such as k-nearest neighbor, naïve Bayes, decision trees, and logistic regression, in terms of accuracy.

A Deep Learning Model to Predict Student Learning Outcomes in LMS Using CNN and LSTM

“…The design goal of the proposed method is predicting students' learning outcomes and their performances in LMS at higher education using two methods: CNN [33] and LSTM [34,35]. By combining two methods; namely, 1) CNN to extract effective features from the data, and 2) LSTM to identify the interdependence of data in time series data, the performance prediction accuracy was improved compared with state-of-the-art methods.…”

“…This makes full use of the time sequence of student data to obtain more reliable predictions. Second, by comparing the CNN-LSTM evaluation indexes with CNN [33] , LSTM [34], RNN [38] and CNN-RNN [39], our method has good prediction accuracy and was better able to predict student performance within our higher education institution. The CNN model is a type of feedforward neural network developed by [36].…”

“…As shown in Fig. 8, we compared CNN-LSTM model with others: CNN [33], LSTM [34], RNN [38] and CNN-RNN [39] in terms of performance evaluation, MAPE and RMSE. It can de seen that our method CNN-LSTM has the smallest prediction error compared with others according to RMSE=39.69 and MAPE=27.56.…”

“…It can de seen that our method CNN-LSTM has the smallest prediction error compared with others according to RMSE=39.69 and MAPE=27.56. Furthermore, F1-score was utilized to evaluate the predication accuracy of our method and compared with other: CNN [33] and LSTM [34].…”

Academic Performance Prediction Model Using Classification Algorithms: Exploring the Potential Factors

“…Thus, many past studies have discussed academic performance from different perspectives, such as knowledge score, grade and performance. They used different criteria or features in measuring academic performance, such as GPA Poudyal et al (2022); Ahmed et al (2018) tests Yagci (2022) quizzes Poudyal et al (2022), course grades Nabil et al (2021) IELTS score (Ghazal & Allah, 2020)and pre-admission test (Mengash, 2020). However, the CGPA and GPA of students are the most common indicators used by researchers in measuring students' academic performance.…”