Predicting Student’s Performance by using Classification Methods

Abstract

Most of the developing countries are facing the problem of ever-rising low-quality population. To convert this low-quality population into a high-quality one, efforts are required to be laid down. These efforts include investment in Research and development and in the education sector. If the people living in an area will be educated then they will be productive for the nation and eventually contribute towards its GDP. The advancement of technology helps educational institutions to turn raw data into actionable insights to achieve desirable results. This study has worked towards prediction models so that student's performance be evaluated timely so that necessary steps be taken in due time to improve their performance. In this study, the record of 265 Computer Science and Engineering students at UIET, MDU Rohtak are used for prediction task by using five algorithms namely Simple Linear Regression, Random Forest, Decision Table, SMOreg, LWL. Algorithms are compared by using WEKA tool. The results showed that out of these five algorithms, Simple Linear Regression gave the best prediction accuracy with the highest correlation coefficient of 0.78, lowest Mean Absolute Error of 4.97 and lowest Root Mean Squared Error of 6.75. This study has laid the foundation in selecting efficient algorithms for predicting the results of students.

Prediction of Number of Participants in The Online Examination System at Universitas Terbuka

“…In determining predictions in data mining, several methods can be used to carry out the process of data analysis, in this study several data mining methods will be used that produce how large the prediction is successful and accurate, namely Linear Regression, Decision Tree, Support Vector Regression and Artificial Neural Network.In several studies, a comparison was made of how accurately the regression algorithm can predict the required results [2], [3]. Some of the use of linear regression as one of the most commonly used methods in making predictions is the study of Prediction of Stock Using Fuzzy Linear Regression [4] and also sales predictions for a bookstore using Simple Linear Regression [5].…”

Improving classification accuracy :The KNN approach

“…Complex structures were easily illustrated by classification trees and random forests in [12] which otherwise would have taken many interaction terms to find using the common regression techniques. Simple linear regression gave the best accuracy among five algorithms compared in [13] to predict student's performance. Reference [14] applied the KNN algorithm among others to predict the performance of students in end semester university examinations.…”

A Business Rule for a B-School using Machine Learning

“…Reference [12] concluded that the decision tree algorithm can be incorporated in the Academic Environment Model to assist lecturers and management to make informed decisions about student performance. In [13], the performance of five algorithms is compared in the prediction of student's performance and results demonstrated that Simple Linear Regression gave the best prediction accuracy.…”