Predicting dropout in online learning environments

Online learning environments became popular in recent years. Due to high attrition rates, the problem of student dropouts immense importance for course designers, and makers. In this paper, we utilized lasso ridge logistic regression create a prediction model dropout on Open University database. We investigated how early can be predicted, why occur. To answer first question, created models eight different time frames, ranging from beginning mid-term. There are two results based definitions dropout. Results show that at AUC is 0.549 0.661 rises 0.681 0.869 By analyzing coefficients, showed demographic features description most important variables prediction, while later activity gains more importance.