CBR-FoX is a Python library designed to provide case-based reasoning explanations for time series prediction models. This approach enhances the transparency and interpretability of machine learning models applied to sequential data.
- Case-Based Reasoning (CBR) Implementation: Utilizes case-based reasoning to enhance explainability in time series predictions.
- Versatile & Adaptable: Supports various types of time series data.
- ML Model Compatibility: Easily integrates with common machine learning models.
- Comprehensible Explanations: Provides clear, human-readable insights into model predictions.
To install CBR-FoX and its dependencies, follow these steps:
# Clone the repository
git clone https://github.com/jerryperezperez/CBR-FoX.git
cd CBR-FoX
# Install required dependencies
pip install -r requirements.txt
Follow these steps to use CBR-FoX in your projects:
Extract the relevant inputs and outputs from your AI model.
from cbr_fox import CBRfoxInstances
cbr_instances = CBRfoxInstances(model_outputs)
from cbr_fox import CBRfoxBuilder
builder = CBRfoxBuilder(cbr_instances)
builder.fit(train_windows, train_targets, target_to_analyze, window_to_predict)
builder.predict(prediction=prediction, num_cases=5)
builder.visualize_pyplot(
fmt='--d',
scatter_params={'s': 50},
xtick_rotation=50,
title='Example Visualization',
xlabel='X-axis',
ylabel='Y-axis'
)
The following diagram illustrates the typical workflow of CBR-FoX, from retrieving AI model outputs to generating visual explanations.
The following diagram represents the core classes and their interactions within the library. The cci_distance
file is utilized when an instance is created using the custom distance metric implemented in this script.
For further details, check out the official repository: CBR-FoX on GitHub.