Smit Topiwala
Board Game Review Prediction
Predicted average board game ratings using Linear Regression and Random Forest Regressor on Kaggle board game data. The project helps identify which game features correlate most strongly with player ratings.
Tools: Python, scikit-learn, pandas, Kaggle dataset
Repo: GitHub
Key Visualizations
Rating distribution
Most ratings cluster around 6.0 (avg: 6.02, std: 1.58), indicating a left-skewed distribution.
Feature correlation matrix
Average rating correlates with average weight and minimum age. Metadata columns like id and name show correlation but carry little predictive value.
Model comparison
Side-by-side comparison of Linear Regression vs. Random Forest predictions.
Key Results
| Model | MSE |
|---|---|
| Linear Regression | 2.08 |
| Random Forest | 1.56 |
Random Forest outperformed Linear Regression, confirming the non-linear relationships in the feature space.