This project contains an overview of main existing methods of automated machine learning:
- Automated feature engineering based on featuretools;
- H2O AutoML;
- AutoKeras;
- SMAC;
Based on dataset, the task was to predict a price of an item in e-commerce. The results are provided here:
| Model Name | R Squared |
|---|---|
| CatBoost Baseline | 0.749 |
| CatBoost with AutoFE | 0.838 |
| H2O AutoML | 0.897 |
| H2O AutoML with AutoFE | 0.892 |
| SMAC | 0.802 |
| AutoKeras on Struct. Data | 0.659 |
Here's a list of other related projects where you can find inspiration for doing AutoML research: