# Multi-split Optimized Bagging Ensemble Model Selection for Multi-class Educational Data Mining

**Authors:** MohammadNoor Injadat, Abdallah Moubayed, Ali Bou Nassif, Abdallah Shami  
**arXiv:** [2006.05031](https://arxiv.org/abs/2006.05031)  
**Date:** 2020-06-09  
**Summary page:** https://www.summarizepaper.com/en/paper/2006.05031v1/

## Key points

- The study focuses on predicting students' academic performance using data mining techniques.
- Analyzing and predicting students' performance can help institutions improve the quality of education and student performance.
- Two different undergraduate datasets from two universities were analyzed to predict students' performance at two stages of course delivery (20% and 50%, respectively).
- A systematic multi-split approach based on Gini index and p-value was adopted to optimize a suitable bagging ensemble learner built from any combination of six potential base machine learning algorithms.
- The posited bagging ensemble models achieve high accuracy for the target group for both datasets.
- Various data mining techniques were used to determine possible factors that may affect the students' final marks.
- Such techniques allow instructors to identify these factors and take necessary steps to improve student performance.
- The study contributes significantly to the field of educational data mining by providing insights into predicting students' academic performance using data mining techniques.

## Summary

The study titled "Multi-split Optimized Bagging Ensemble Model Selection for Multi-class Educational Data Mining" focuses on predicting students' academic performance using data mining techniques. The authors note that many institutions are interested in improving the quality of education and student performance, and analyzing and predicting students' performance can help achieve this goal. To this end, the authors analyze two different undergraduate datasets from two universities to predict students' performance at two stages of course delivery (20% and 50%, respectively). The study adopts a systematic multi-split approach based on Gini index and p-value to optimize a suitable bagging ensemble learner built from any combination of six potential base machine learning algorithms. This approach allows for properly choosing appropriate machine learning algorithms to use as well as optimizing their parameters. The authors show through experimental results that the posited bagging ensemble models achieve high accuracy for the target group for both datasets. Furthermore, the study aims to determine possible factors that may affect the students' final marks by analyzing various data mining techniques. The authors emphasize that such techniques allow instructors to identify these factors and take necessary steps to improve student performance. Overall, this study contributes significantly to the field of educational data mining by providing insights into predicting students' academic performance using data mining techniques. The findings can be highly beneficial for institutions looking to improve education quality and student performance.