engleski

Setting Up Automated Programming Assessment System for Higher Education Database Course

The need for decreasing the workload among educators, timely feedback and accuracy and consistency on the grading results are the common reasons that motivate the development of Automated Programming Assessment Systems (APAS). Using our newly deployed APAS on Databases course we have evaluated students’ knowledge using a “little and often” pattern and employed gathered data to predict students’ performance on the course. The course’s pass percentage and students’ grades were predicted using several data mining techniques. Besides internal data gathered during course’s execution, additional external variables (like grade point average) were considered for the data mining models. Our analysis shows that the accuracy of prediction is not highly affected if external variables are unknown. The pass percentage accuracy predictions is sufficiently high, especially after the half of semester (~80%), which allows for proactive approach towards the students we believe will fail the course. Other than that, we came up with few valuable insights into the structure and content of assignments that shall be applied in the next course cycle, in accordance with our intention to use course’s data to improve the teaching and optimize the staff’s man-hours.

Educational Data Mining ; Academic Performance ; Prediction ; Automated Assessment

nije evidentirano