NextGen Learning: Hybrid Mode Prediction with Machine Learning
Chaman Verma · 2024
This paper presents the machine learning model to predict whether hybrid learning will be a sustainable solution. A binary classification problem to identify the hybrid learning mode has been solved. The primary samples are trained and tested with the help of the Random Forest (RAF) algorithm with 86% accuracy. The learning curve of the yellowbrick library proved an insignificant difference between the test and training accuracy and a sufficient number of samples trained for the RAF model. The calibration curve found the highest probability of 90% occurrence of positives. The RAF model gained high precision, recall, and a f1-score of 0.86. At threshold 50, precision is 0.66, f-score is 0.77, accuracy is 0.69, and recall is 0.96. The RAF model's output is also validated and explained using SHAP (SHapley Additive exPlanations), and novel significant features are recommended.