A Novel Approach in Determining the Reasons of Student Attrition based on Enhanced Genetic Algorithm with Cross-Average Crossover Operator

Prediction helps organizations in their decision-making activities for the improvement of the different functions, services, and even their income. The C4.5 algorithm is the most commonly used prediction algorithm having difficulty in choosing attributes that lead to low prediction accuracy. Hence, integration of k-means segmentation for feature selection has been done showing an improvement of prediction accuracy. However, the accuracy result is still not good enough for a prediction model. Thus, there is a need to develop a new strategy to increase the accuracy to an acceptable level. In the present study, Genetic Algorithm (GA) is enhanced and integrated into the prediction model. A new crossover mechanism called Cross-Average Crossover (CAX) is introduced to address the problem of the original GA that can lead to not so accurate prediction results. The student's reasons for attrition in an educational institution was used as the dataset to determine the accuracy of the newly developed model having GA with CAX as against to a model having unenhanced GA. Results showed that an enhanced GA outperformed the unenhanced GA in terms of variable optimization. Further, a prediction model having an improved GA yields a more accurate prediction results as against to a model having unenhanced GA