Implementation of Ensemble Learning to Predict Learner’s Attainment—A Random Forest Classifier

Random forest classifier is used both for the classification of data and to perform regression analysis. It performs a high level of accuracy in making predictions by constructing a forest of decision trees and taking the average output of all the trees constructed. The learner's attainment is usually a grade or score attained at the end after completing the academic education. It is a learner's progress over a period of time from his enrollment to any institution till the end of the entire course. It is the difference between the attainment the learner attains in the first year that is Key Stage 1 and the last year which is Key Stage 4 of four years of graduation. Initially at Key Stage 1, only two features academic and physical developments are considered for study. A model is developed to implement a random forest classifier that calculates the average score for the Key Stage 1 that draws accuracy and predictions. The accuracy comes out to be 74.40% which is idealistic. To derive more realistic accuracy and to gain the best attainment by learners, more Social, Moral, and Physical development activities are conducted for Key Stages 2, 3, and 4, respectively. The learner’s participation is observed and average score is recorded. Further, the same classifier is implemented on the learner's Key Stage 4 data, and the accuracy comes out to be 96%. The comparative study of other metrics such as F1-score, precision, and recall is performed for both levels of implementations. The model shows difference in accuracy of classifier implemented at Key Stages 1 and 4. The confusion matrix shows difference in true positive values for AVERAGE, GOOD, and EXCELLENT classes, respectively, that except few rest all the learners have improved in all the features with the GOOD and EXCELLENT attainments.