This repository contains a solution to the multiclass classification problem on the Iris dataset using the CatBoostClassifier model.
- python3 train.py for model training
- python3 infer.py to get predictions on the test dataset, dataset with obtained predictions will be stored in data > predicts
Go to file train.yaml located in the "configs" directory and modify model > optimizer_parameters section
Go to file train.yaml located in the "configs" directory and modify model > custom metrics section (please refer to CatBoost documentation to get available metrics)
- Datasets are stored in dvc using Google Drive as backend
- All files created during training / inferring are also saved to dvc
- Training parameters and metrics are logged using MLFlow
- Start and end of training / inferring steps are logged to console