| Algorithm | Optimized Hyperparameters |
|---|---|
| Neural Network | Activation function: sigmoid, Batch size: 512, decay:1.47143e-05, epochs: 83, Hidden layers: 1, Hidden nodes: 170, l1_reg: 2.89385e-07, l2_reg: 1.001992e-06, learning_rate: 0.0200339, pinball_tau: 0.256, Loss function: pinball loss |
| Gradient Boosting | criterion: 'friedman_mse', alpha': 0.208, max_depth: 16, max_features:0.439, min_samples_leaf: 13, min_samples_split: 6, n_estimators: 237, loss: Quantile |
| Random Forest | criterion': 'squared_error', max_depth: 19, max_features: 0.599, min_samples_leaf: 18, min_samples_split: 15, n_estimators: 283, quantile: 0.088 |