00098
PREDICTION OF STUDENT SUCCESS RATE USING NAIVE BAYES

Saturday, February 18, 2017
Exhibit Hall (Hynes Convention Center)
Kshitija Karkar, CSU CHICO, CHICO, CA
Educational institutions now-a-days have become more committed to deliver better student performance. Academic performance is measured in terms of grades for a student. This is a research project that implements a suitable prediction algorithm for calculating future grades. The project uses data mining tool called RapidMiner to predict the success rate of a student based on certain factors that includes previous grades, study time and attendance by applying the Naïve Bayes probability theorem. A model on final grades is created using Naïve Bayes and training dataset. This model is then applied on the testing dataset to predict final grades from the testing dataset. Accuracy of this model is calculated by cross checking final grade values obtained by applying the model and the original values to get approximately 85% similarity. The project also calculates effect of study time and attendance on grades. A comparison between predicted and actual grades gives a clear view of the efficiency of this model.