Volume 18, No. 6, 2021

Student Peformance Prediction Using Feature Imbalance Aware Xgboost Algorithm


Shashi Rekha H , Dr. Chetana Prakash

Abstract

in the current situation of the pandemic, most of the schools and institutions have changed their mode of teaching to the online mode, hence there is a sudden increase in the e-Learning Systems. Due to the e-Learning Systems, many students are facing issues connecting to these platforms. Some of the issues include no electricity, no proper internet connection, etc. Hence, there is a slight decrease in the student’s performance. Furthermore, some of the institutions are trying to improve the student performance and quality of education in the e-Learning Systems using Data Mining (DM) employed Machine Learning (ML) Technique. These techniques are used to analyze the student activity such as session time, login time, time spent in the eLearning Systems, etc., and then predict the performance of the student. Some of the studies have shown that the Machine Learning-Based techniques give a correct result only when the data is balanced. Hence it is required to choose the correct Machine Learning algorithm according to the data. Most of the existing Student Performance Prediction Techniques have designed their models by combining various Machine Leaning Algorithms to choose the best model according to the data. Furthermore, these techniques have not incorporated the feature importance to predict the performance of the student. Hence, this results in poor performance mostly for the multi-label classification. Thus, this paper gives a model using the XGBoost (XGB) Algorithm, named Feature Imbalance Aware XGB (FIA-XGB). The FIA-XGB uses the effective cross-validation technique to learn the correlation between the features and increase the performance of the model efficiently. The results show better performance in terms of prediction accuracy when compared with the existing Machine Learning Student Performance Prediction models.


Pages: 2228-2238

Keywords: Handover execution, Heterogeneous wireless network, Radio access technology selection, Machine learning.

Full Text