A prediction-based curriculum analysis using the modified artificial Bee Colony Algorithm

Abstract

Due to the vast amount of students' information and the need of quick retrieval, establishing databases is one of the top lists of the IT infrastructure in learning institutions. However, most of these institutions do not utilize them for knowledge discovery which can aid in informed decision-making, investigation of teaching and learning outcomes, and development of prediction models in particular. Prediction models have been utilized in almost all areas and improving the accuracy of the model is sought- after this study. Thus, the study presents a Scoutless Rule-driven binary Artificial Bee Colony (SRABC) as a searching strategy to enhance the accuracy of the prediction model for curriculum analysis. Experimental verification revealed that SRABC paired with K-Nearest Neighbor (KNN) increases the prediction accuracy from 94.14% to 97.59% than paired with Support Vector Machine (SVM) and Logistic Regression (LR). SRABC is efficient in selecting 14 out of 60 variables through majority voting scheme using the data of the BSIT students of Davao Del Norte State College (DNSC), Davao del Norte, Philippines.