The application of formative e-assessment data in final exam results modeling using neural networks

After introducing e-assessment into laboratory and seminar teaching of physics in a biomedical university study program the improved final practical exam passing rate was observed. The final practical exam passing rate increased from approximately 60% to more than 80%. Encouraged by these results, in this paper we will try to show a correlation between the results of e-assessments collected during continuous monitoring of student performance and their success in final tests using neural networks as modeling method. The models will be built using various sets of data collected during three academic years. Aside from the impact on the final practical exam passing rate, the models should help us in dealing with the missing data problems. Some data are missing due to technical problems and some due to errors in student handling of e-application. In this paper we have observed data collected by two types of e-assessment. The first type of e-assessment is carried out during laboratory exercises and it influences the formation of the final grade. The second type of e-assessment is performed during seminars and has no impact on the final grade.