The Effective Learning Approach to ICT-TPACK and Prediction of the Academic Performance of Students Based on Machine Learning Techniques

Abstract

Objectives: To analyze and compare different techniques of machine learning for the student’s academic performance to find an effective learning environment (ICT-TPACK). Methods: A descriptive study design was adopted among 3000 BCA and B.Sc. (computer science) students from various locations at Bharathidasan University affiliated colleges in the Karur region, Tamilnadu, India. The data was collected through internal marks from June 2019 to April 2021. The internal mark or grades are calculated for the time of submission of the assignment, attendance, evolution of internal examination answer script, MCQ test, and student’s study seminar. Findings: This research paper was analyzed for ANN, LSTM, MLP, and our proposed methods (IP-LSTM). The proposed algorithms (IP-LSTM) can better reflect the academic performance of the students; With the LSTM method, some information can be forgotten. Our proposed method achieves 91% accuracy and 94% recall, which outperforms the other methods. Novelty: Introducing a new algorithm in ML-called IP-LSTM for knowing the effective learning techniques to improve students’ academic performance to create an effective environment for learning.