Synthetic Data Generation from Sample Data Analysis and Modeling

To efficiently tackle the challenge of responsible gaming in wager-based games, it is necessary to be able to understand the behavior of players. A first step in this is to identify playstyles. To identify playstyles, a set of features are derived from session level data. The resulting features are then clustered using a clustering algorithm. Instead of extracting a defined set of features, we want to learn the features and then cluster the resulting representations to identify playstyles. It is difficult to provide a simple interpretation for the resulting clusters and as such we aim that given a dataset and its resulting clusters, to develop an algorithm to explain the resulting clusters. Once we have fundamentally understood the features and clusters, we will develop models of players that can then be used to generate synthetic player data. Synthetic player data will support exploration of artificial intelligence and machine learning solutions in the domain whereby player data is needed to support training. In this case, synthetic data can help achieve the goals desired however without having privacy and security concerns.

Faculty Supervisor:

Kenneth Kent;Suprio Ray

Student:

Partner:

IGT

Discipline:

Computer science

Sector:

Arts, entertainment and recreation; Information and cultural industries

University:

University of New Brunswick

Program: