Modelling

[Spencer-Weston]

Baseline: Linear Regression

Other Regressions (Matt)

Improve out of sample prediction accuracy by using regularization to avoid overfitting on the training set.

• Lasso regression
• Ridge regression

Random Forest/Regression Tree (Weijia)

• regression tree
• random forest
• Boosting/ XGboost

Neural Network

Spencer fiddles around with neural networks until one works.

[Spencer-Weston]

Performance Metrics: Win/Loss: Accuracy, F-Score Margin of Victory: R^2 and MSE