Automatic Machine Learning
Using H2O AutoML Workflow
- Prep work: I load my data, point out the target column “Life expectancy”, and impute missing values.
- Let H2O do its magic: AutoML takes over, and benchmarks hundreds of different models using cross-validation. Here are the top performers:
| model_id | rmse | mse | mae | rmsle | mean_residual_deviance |
|---|---|---|---|---|---|
| StackedEnsemble_AllModels_5_AutoML_8_20240523_123256 | 1.5141 | 2.2926 | 0.7862 | 0.0235 | 2.2926 |
| StackedEnsemble_AllModels_4_AutoML_8_20240523_123256 | 1.5210 | 2.3134 | 0.8363 | 0.0236 | 2.3134 |
| StackedEnsemble_AllModels_3_AutoML_8_20240523_123256 | 1.5612 | 2.4374 | 0.8530 | 0.0241 | 2.4374 |
| GBM_grid_1_AutoML_8_20240523_123256_model_11 | 1.5655 | 2.4509 | 0.8601 | 0.0242 | 2.4509 |
Explainability: Models can be used more effectively if you know why they perform well. Here are some explainability charts for top-performing models.
Put it to work: I export the chosen model and use it to make predictions on new data.