In this paper, we investigate predicting students at risk of dropping out using academic data from an online study program offered at two different universities. In this program, most students work full-time and thus study part-time. We use the algorithms decision tree and logistic regression, which have given good results in other works and are interpretable. The decision trees or the coefficients produced by logistic regression can be shared with stakeholders to identify courses that impact dropout the most and that could be considered for additional support. Our results are comparable to the results that others have obtained when considering face-to-face study programs. We propose a preliminary approach to identify courses that impact the prediction the most. Our results also show that merging the data from the two universities is not helpful for the prediction. This might be due to differences in the regulations that impact how students study.