AutoCarver

Create a Python-based data preprocessing tool called 'FeatureDiscretizer' that leverages the 'AutoCarver' package to automatically discretize numerical features into categorical bins, optimizing these bins for target association. This tool will serve as a pre-processing step for machine learning models, helping to improve model performance by transforming continuous variables into discrete ones that may capture more meaningful patterns. The project should include the following steps and features:

1. **Project Setup**: Initialize a new Python project with necessary dependencies including 'AutoCarver', pandas, numpy, and scikit-learn.
2. **Data Loading Module**: Develop a module that allows users to load their dataset from various formats like CSV, Excel, or SQL databases. Ensure the module can handle missing values and outliers appropriately.
3. **Feature Discretization Module**: Implement the core functionality using 'AutoCarver'. This module should take numerical columns from the input dataset and apply 'AutoCarver' to transform them into categorical bins. Each bin should be optimized to maximize its association with the target variable, ensuring that the discretized features are as informative as possible for subsequent modeling tasks.
4. **Visualization Module**: Integrate matplotlib or seaborn for visualizing the original distribution of each numerical feature alongside its discretized counterpart. This will help users understand the transformation applied and verify if it makes sense for their specific use case.
5. **Model Evaluation Module**: After discretization, use scikit-learn to train a simple classifier or regressor on the transformed data and compare its performance against a model trained on the raw data. This comparison should be done using cross-validation to ensure robust evaluation.
6. **User Interface**: Design a simple command-line interface (CLI) that guides users through the process of loading their data, selecting features for discretization, applying the transformation, and evaluating the results.
7. **Documentation & Testing**: Provide comprehensive documentation detailing how to install and use the 'FeatureDiscretizer' tool, along with examples. Also, write unit tests to validate the correctness of the implementation and edge cases.

By completing this project, you'll create a powerful yet user-friendly tool that demonstrates the practical benefits of using 'AutoCarver' for data preprocessing, potentially enhancing the predictive power of downstream machine learning models.