Vers l'identification des profils des joueurs à travers l'analyse des données du jeu basée sur le modèle de régression et le regroupement

Abstract

The customization of serious games is an important factor in their success and effectiveness as motivating and attractive tools in several fields such as education and health. However, to achieve this purpose, the player profile will be determined through the analysis of game data. It is a large volume of data, including characteristics and participation of the player, that poses challenges about choice of the analysis techniques and data to consider. Our study contributes to a better understanding of the player progression in a virtual environment through the data mining of serious game "Science en jeu". It aims at identifying relevant data and appropriate data mining methods as well as the deduction of player profile features. For that purpose, we used two methods namely multiple linear regression and Clustering (Kmeans). The first method showed that the number of access to the game, quests visited and advantages used significantly contribute to the scores and the duration of a game, while the second method revealed three forms of participating players: beginner, intermediate and advanced, who interact with the game according to their experiences. These results provide us with a first reading of players’ profiles which would be enriched with data assessment tests provided by experts (for example, teachers or therapists). Also, the test methods for classification and testing of the game in the context of education or health should be considered in future studies.

Date	29 Sept 2014
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Nadjia Kara (Supervisor) & Neila Mezghani (Co-supervisor)