-------------
Released 2024-05-25
- Added support for binary classification model evaluation.
- Allows specifying a threshold for model output probabilities.
- If no threshold is provided, FlameAI will automatically search for the optimal
threshold based on an objective function (e.g., maximizing precision).
- Added support for linear regression model evaluation.
- Added support for data preprocessing.
- Convert categorical labels into numeric format.
- Calculate the scaling factor for an imbalanced dataset.
- Implement a simple DataLoader.
- Added support for plotting ROC curves and confusion matrices.
- Added support for a command-line tool.
- Check if Python, Hive and PyTorch are installed in the current env.
- Hive command-line tool to execute an HQL file and save the result to a CSV file
with the same name as HQL file.
- Run nox tests using GitHub Actions.
- Added some examples.
- Removed several modules from ``__all__`` to fix the issue of slow loading of FlameAI.
- Removed torch from ``pyproject.toml``'s dependencies.