Board Game Data Analysis
Statistical analysis and modeling of Board Game Geek data to understand trends in the production quantity and quality of Board Games over time. The project uses probabilistic models to uncover patterns in the board game industry.
Modeling Quantity
- Modeled annual game production as a Poisson process with time-varying rate
- Used exponential function for underlying rate: λ(t) = exp(at + b)
- Fitted via stochastic search with R² = 0.983
Modeling Quality
- Modeled proportion of high-scoring games as binomial with changing probability
- Probability modeled as exponential function over time
- Achieved R² = 0.868 for probabilities, R² = 0.991 for counts
Hypothesis Testing:
- Themes: Warfare games scored highest; knowledge-based games scored lowest
- Mechanics: Games discouraging player interaction performed best
- Funding: Kickstarter-funded games significantly outperformed non-Kickstarter games
The analysis reveals clear temporal trends in both the quantity and quality of board games produced, with significant differences across themes, mechanics, and funding sources.