Evaluating Performance Of Identifying At - Risk Students And Learning Achievement Model using Accuracy And F-measure by Comparing Logistic Regression, Generalized Linear Model And Gradient Boost Machine Algorithm

• Published in: 2022 International Conference for Advancement in Technology (ICONAT) pp. 1 - 7
• Main Authors: Harini, K., Rekha, K K. Sashi
• Format: Conference Proceeding
• Language: English
• Published: IEEE 21.01.2022
• Subjects: Accuracy; At-risk Students; Education; Effective teaching; F-Measure; GBM algorithm; GLM Algorithm; Learning Achievement; Logistics; Machine Learning; Novel Sigmoid Logistic Regression; Prediction; Prediction algorithms; Predictive models; Probability; Significantly; Statistical Probabilities
• DOI: 10.1109/ICONAT53423.2022.9725848

Aim: The objective of the study is to improve the Accuracy and F-Measure percentage of At - risk students Model on the basis of High withdrawal and Failure rate and learning Achievement Model based on low performance and less motivational status. Materials and Methods: Logistic Regression with sample size = 20, Generalised Linear Model (GLM) with sample size = 20 and Gradient Boost Machine algorithm (GBM) with sample size = 20 was executed a different number of times for predicting accuracy and F-Measure percentage of At - risk students Model and Learning Achievement Model. For improving the Accuracy and F-measure percentage the Novel sigmoid function was used in Logistic Regression which helps in mapping the prediction to probabilities. Results and Discussion: Logistic Regression achieved significantly better Accuracy (95.61%), F-Measure (93.65%) compared to GLM Accuracy (92.97%), F-Measure (89.45%) in At-Risk students Model and in Learning Achievement Model Logistic Regression has better accuracy (96.37%), F-Measure (95.59%) when compared to GBM accuracy (91.73%), F-Measure (90.85%). There was a statistical significance between Logistic regression, GLM and GBM (p = 0.000) (p < 0.05). Conclusion: In this study it is found that the Logistic Regression algorithm performed better than the GLM algorithm and GBM Algorithm for predicting accuracy, F-Measure percentage of At - risk students Model and Learning Achievement Model using Novel Sigmoid function.