This document discusses using data mining classification approaches and genetic algorithms to predict student grades in an online educational system. It proposes using student data like homework submissions and web usage to classify them and predict their final grades. A genetic algorithm is used to optimize combinations of different classifiers like k-nearest neighbor and k-means on a dataset from an online learning system called LON-CAPA. The performance of several classification algorithms is evaluated to predict student grades based on their online activities. Sample student and faculty datasets are also presented.