A novel framework for envisaging a learner's performance using decision trees and genetic algorithm

In order to improve the overall performance of an institution, individual performances must be looked into. Hence it is useful for educational institutions to analyze learners' performances to identify the areas of weakness to guide their students to a better future. In this paper, an algorithm is proposed for predicting a learner's performance using decision trees and genetic algorithm. Id3 algorithm is used to create multiple decision trees, each of which predicts the performance of a student based on a different feature set. Since each decision tree provides us with an insight to the probable performance of each student; and different trees give different results, we are not only able to predict the performance but also identify areas or features that are responsible for the predicted result. For higher accuracy of the obtained results, genetic algorithm is also incorporated. The genetic algorithm is implemented on the n-ary trees, by calculating the fitness of each tree and applying crossover operations to obtain multiple generations, each contributing to creating trees with a better fitness as the generations increase, and finally resulting in the decision tree with the best accuracy. The results so obtained are quite encouraging.