info-circle
student.uva.nl is vernieuwd. Lees meer...
Welke opleiding volg je?
Lezing

Presentation Master's Thesis - Lisette Sibbald - Psychological Research Method

Laatst gewijzigd op 26-07-2022
Scalable, Robust and Maintainable Prediction of European Credit Transfer and Accumulation System Points
Toon informatie specifiek voor jouw opleiding
Nu zie je algemene informatie op deze site. Kies je opleiding om ook informatie te zien die specifiek voor jouw opleiding geldt, zoals deadlines, regelingen en contactgegevens.
Welke opleiding volg je?
Startdatum
04-08-2022 11:00
Einddatum
04-08-2022 12:00
Locatie

Roeterseilandcampus REC GS.04

To facilitate financial planning of Societal and Behavioral Faculty of the University of Amsterdam (UvA), the UvA Advanced Analytics team recently introduced a linear model that predicts European Credit Transfer and Accumulation System points (ECTS) at the end of the academic year per student based on a model trained on the previous academic year. However, this model does not perform as desired since the goal is to have an average prediction error below 5 percent. The origin of this high error is thought to be related to issues with the independent variable (IV) that has the most predictive value, ECTS obtained per student up until the month of prediction (CECTS). This IV shows different distributions in different years. This difference is quantified and shown to correlate with prediction error. In this study it is shown that a decision tree is a more natural model for this prediction and that performance is increased in situations of large difference in the distribution of this IV between years.