Background: Technological innovations such as Learning Management Systems (LMS) are becoming more and more prevalent in the learning environments of students. Distilling and acting on knowledge gathered from these systems, the field known as learning analytics, allows educators to hone their craft and support students more effectively by providing timely interventions. Objectives: While most learning analytics studies focus on using LMS data to predict performance, this study instead predicts students' self-reported time-management skills using trace data from the Canvas LMS. This is done for courses at one Dutch technical university with in total 462 students. Methods: Linear regression and multi-level regression models are constructed using both theory and findings from previous research. The predictions made by these models are compared to previously filled in questionnaire data to validate the results. Results: Our results show that models can be constructed and time-management can be predicted for individual courses. Furthermore, there are several predictors that are significant in multiple models. However, these models and predictions are not immediately transferable to other courses. Conclusions: The study therefore emphasizes the need for further research, using multiple sources of data or more theoretically grounded predictors, to investigate the extent of the portability issues with these predictive models. Despite this we were able to predict the students' self-reported time-management skills in multiple different courses using Learning Analytics, and managed to identify multiple consistently predictive trace data variables.