Comparing Different Resampling Methods in Predicting Students’ Performance Using Machine Learning Techniques

Abstract

In today's world, due to the advancement of technology, predicting the students' performance is among the most beneficial and essential research topics. Data Mining is extremely helpful in the field of education, especially for analyzing students' performance. It is a fact that predicting the students' performance has become a severe challenge because of the imbalanced datasets in this field, and there is not any comparison among different resampling methods. This paper attempts to compare various resampling techniques such as Borderline SMOTE, Random Over Sampler, SMOTE, SMOTE-ENN, SVM-SMOTE, and SMOTE-Tomek to handle the imbalanced data problem while predicting students' performance using two different datasets. Moreover, the difference between multiclass and binary classification, and structures of the features are examined. To be able to check the performance of the resampling methods better in solving the imbalanced problem, this paper uses various machine learning classifiers including Random Forest, K-Nearest-Neighbor, Artificial Neural Network, XG-boost, Support Vector Machine (Radial Basis Function), Decision Tree, Logistic Regression, and Naïve Bayes. Furthermore, the Random hold-out and Shuffle 5-fold cross-validation methods are used as model validation techniques. The achieved results using different evaluation metrics indicate that fewer numbers of classes and nominal features will lead models to better performance. Also, classifiers do not perform well with imbalanced data, so solving this problem is necessary. The performance of classifiers is improved using balanced datasets. Additionally, the results of the Friedman test, which is a statistical significance test, confirm that the SVM-SMOTE is more efficient than the other resampling methods. Moreover, The Random Forest classifier has achieved the best result among all other models while using SVM-SMOTE as a resampling method. INDEX TERMS Classification, data mining, educational data mining, imbalanced data problem, machine learning, resampling methods, statistical analysis.

Separating emission and meteorological contribution to PM<sub>2.5</sub> trends over East China during 2000–2018

“…We assimilated the daily PM2.5 measurements, WRF/CMAQ simulations, satellite aerosol optical depth (AOD) from Aqua and Terra MODIS Level 2 products (https://ladsweb.modaps.eosdis.nasa.gov/), meteorological parameters from the Modern-Era Retrospective analysis for Research and Applications Version 2 (MERRA-2) (Randles et al, 2017;Buchard et al, 2017), elevation data from the Global Digital Elevation Model (GDEM) (https://earthexplorer.usgs.gov/), gridded population distributions (Xiao et al, 2021b), and land cover classification data (http://data.ess.tsinghua.edu.cn) (Gong et al, 2019a; to train the PM2.5 prediction model and predicted PM2.5 concentrations during 2000-2018. The detailed data collection and processing methods were summarized in Appendix A.…”

“…A total of 3.9% of the daily data were assigned as high-pollution. Previous studies reported that balancing training data with SMOTE improved the classifiers' performance (Ghorbani and Ghousi, 2020;Saputra and Suharjito, 2019). Thus, we applied the SMOTE algorithm that oversampled the high-pollution data (the minority) by artificially generated new synthetic samples along the line between the high-pollution data and their selected nearest neighbors (Chawla et al, 2002;Chawla et al, 2003).…”

“…We calibrated the gridded population distribution data from the LandScan Global (https://www.worldpop.org/) at the county level with the total population reported in China City Yearbooks. These calibrated gridded population data were fused to better characterize the population distribution across China (Xiao et al, 2020).The land cover classification data of urban and rural regions at a 30-m resolution for 2000-2017 were downloaded from http://data.ess.tsinghua.edu.cn (Gong et al, 2019a;. The fraction of urban/rural region at the 30-m resolution was averaged according to the modelling grid.…”

A New Hybrid Predictive Model to Predict the Early Mortality Risk in Intensive Care Units on a Highly Imbalanced Dataset

“…This method really helps the predictive models to present excellent and trustable performance [39,40]. In 2020, Ghorbani and Ghousi [41] compared various resampling techniques such as Borderline SMOTE, Random Over Sampler, SMOTE, SMOTE-ENN, SVM-SMOTE, and SMOTE-Tomek to handle the imbalanced data problem while using different datasets. The results of their research work reveal that the SVM-SMOTE is more efficient than the other resampling methods, and this method improves the performance of classifiers.…”

Study Of Students’ Performance Prediction Models Using Machine Learning

“…There are a number of variables that influence the decision to forecast and track student success in educational institutions. Educational institutions' decision-makers use these variables to create or build strategies to enhance and track students' academic performance [5].…”