This thesis describes an approach and its implementation for the income prediction of a casino using machine learning. First, a short introduction to the field of machine learning and some of its basics is given. Afterward, the thesis describes the related work and some concepts done before. The data used from the casino of the experiments is anylzed in detail und correlations between the features are shown, for example the linear correlation between the games played and the games won. Additionally the data is extended by additional data, like the weather temperature and some commodity trading values like gold and oil. A preprocessing chain that includes lagging and scaling the features preprocesses the data set to prepare it for machine learning experiments. The step for lagging includes lagging the features from one to seven days, one month, one-quarter of a year and one year. With this preprocessed data set experiments with three different machine learning models is done. These experiments include a neural network and an RBF, polynomial, linear and sigmoid kernel SVM. To create a baseline for the predictions with this models this thesis creates a dump predictor via averaging the same three weekdays to predict the next one. For example, the last three Saturdays are used to predict the next Saturday. After training and testing the different models, this thesis comes to the conclusion that only the linear and the sigmoid kernel SVM have a better prediction than the dump predictor with the sigmoid kernel being the best one. This conclusion also shows that this thesis can predict the income for a casino to some extend but more experiments and data is needed to improve those predictions further.
  author = {Schett, Matthias},
  title = {Predicting Casino Business Figures with Machine Learning:
  		  a Case Study},
  month = jun,
  year = {2017},
  note = { {Supervision in corporation with Stephan Dreiseitl,
  		  Department of Software Engineering}},
  institution = {Department of Embedded Systems Design, School of
  		  Informatics, Communication and Media, University of Applied
  		  Sciences Upper Austria},
  type = {Master's Thesis}