Machine Learning
Machine Learning Algorithms
Different Machine Learning Algorithms were used. A complete list can be found below:
- Lasso Regression
- Linear Regression
- Random Forest
- Extra-Trees
- XGBoost
- LightGbm
- ANN
- Random Forest with Message Passing
- XGBoost with Message Passing
- MLP with Message Passing
- Graph Attention Networks
Machine Learning Pipeline
For the training of the models the following techniques were used:
- Grid Search or Halve Grid Search
- 4 Folds Cross Validation
Evaluation
The metrics used for the evaluation are: MAE, MAPE
Results
| Model | MAE | MAPE |
|---|---|---|
| Lasso Regression | 10.889 | 0.261 |
| Linear Regression | 10.890 | 0.261 |
| Random Forest Regressor | 9.692 | 0.223 |
| Extra Trees Regressor | 10.519 | 0.238 |
| LightGBM | 9.447 | 0.214 |
| XGBoost | 9.136 | 0.209 |
| Message Passing RF | 10.248 | 0.241 |
| Message Passing XGBoost | 10.467 | 0.236 |
| Message MLP | 12.415 | 0.291 |
| Graph Attention Network | 10.659 | 0.236 |