MSc Project (30 ECTS) Defence - Department of Computer Science - Thorsten Alexander Roloff
Title: Gamified Education: Building Recommendation System for an Educational Game Platform for Elementary School Children
Date and time: 26th of May at 14:00 Room: M 105
Abstract:
This thesis discusses the data and various models for building a recommendation system for educational game platform specifically targeted towards elementary school level. This project was done in collaboration with a company with a focus on educational games for elementary school children and the data used in this thesis is from, as well as the recommendation system being for, their online education platform. The data received was explored and analysed. The recommendation system uses data from a single play session to recommend other educational games. Different methods were used to generate recommendations. The methods used include XGBoost, Random Forest, and a Deep neural network. The models where evaluated based on accuracy using root-mean-squared-error. The Shap framework was used to further examine the models to see which features from the input data set where the most important. The features found to be most important to the model fit with prior assumptions on which features were thought to be most important. The results seem promising and suited to be used as a recommendation system.
Committee members:
-
María Óskarsdóttir, Associate Professor, Reykjavík University, Supervisor
-
Anna Sigríður Islind, Associate Professor, Reykjavík University, Supervisor
-
Nidia Guadalupe Lopez Flores, Ph.D. Researcher, Reykjavík University, Supervisor
-
Yngvi Björnsson, Professor, Reykjavík University, Examiner
-
Sebastián Maldonado Alarcón, Professor, University of Chile, Chile, Examiner
