This journal review summarizes a research article that implemented an information gain feature selection technique in machine learning classification algorithms to predict students' academic performance. The research used data from 395 Portuguese students to identify factors influencing final grades. It found that applying the information gain technique improved the accuracy of decision tree, random forest, ANN, SVM, and naive Bayes classifiers by reducing irrelevant features. The technique selected the top ten features and achieved the best results, demonstrating the highest accuracy scores for all machine learning models used in predicting academic performance.
This journal review summarizes a research article that implemented an information gain feature selection technique in machine learning classification algorithms to predict students' academic performance. The research used data from 395 Portuguese students to identify factors influencing final grades. It found that applying the information gain technique improved the accuracy of decision tree, random forest, ANN, SVM, and naive Bayes classifiers by reducing irrelevant features. The technique selected the top ten features and achieved the best results, demonstrating the highest accuracy scores for all machine learning models used in predicting academic performance.
This journal review summarizes a research article that implemented an information gain feature selection technique in machine learning classification algorithms to predict students' academic performance. The research used data from 395 Portuguese students to identify factors influencing final grades. It found that applying the information gain technique improved the accuracy of decision tree, random forest, ANN, SVM, and naive Bayes classifiers by reducing irrelevant features. The technique selected the top ten features and achieved the best results, demonstrating the highest accuracy scores for all machine learning models used in predicting academic performance.
Student ID : 18120084 Program : IT (Information Technology) Course : Machine Learning
Title Implementation of Information
Gain Feature Selection Technique in Machine Learning Classification Algorithm for Predicting Students' Academic Performance Journal 2016 National Seminar on Information Technology and Multimedia at STMIK AMIKOM Yogyakarta Volume & Page Vol. 2.9, Page. 55-60 Year 2016 Writer Betha Nurina Sari Reviewer Michael Ngefak (18120084) Date March 19th 2020 Research Goal The main goal of this research is to identify relevant factors influencing students' academic performance by implementing the information gain feature selection technique in a machine learning classification algorithm Research Subject The subjects of this research are students in the Alentejo region of Portugal, with a total of 395 data used, employing machine learning classification algorithms including Decision Tree, Random Forest, ANN, SVM, and Naïve Bayes Research Method The method used in this research is a classification method where, from the collected data, factors influencing the final academic grades of students are identified Research Results Overall, the results of this research provide several pieces of evidence that the implementation of the information gain feature selection technique can influence the accuracy of machine learning classification algorithms (J48, Random Forest, MLP, SVM (SMO), and Naïve Bayes) in predicting students' academic performance in the subject of Mathematics. The results of this research indicate that implementing the feature selection technique can enhance accuracy as irrelevant features to the classification target have been reduced. The information gain feature selection technique, which selects the top ten features, yields the best results in this study, as demonstrated by the highest accuracy achieved by all machine learning classification algorithms used for prediction Research Power The strength of this research is the tool used, which is data classification, and it is quite user-friendly for the research subjects, so data collection does not require a long time as in qualitative methods