CBR-FoX: A case-based reasoning software tool for auditing time series predictions

CBR-FoX is an open-source Python software tool that provides explainable artificial intelligence capabilities for time series forecasting across multiple scientific domains. The tool implements a novel Case-Based Reasoning approach that generates explanatory case bases through sliding-window techniques, enabling researchers to understand and validate model predictions by retrieving historically similar situations. CBR-FoX integrates a flexible adapter mechanism that allows the use of established time series similarity metrics from external libraries such as sktime, alongside its own novel Combined Correlation Index, with signal processing capabilities using a locally weighted scatterplot smoothing filter to enhance explanation quality. The software is available under the European Union Public License 1.2 at https://doi.org/10.5281/zenodo.17381828 , tested on Python 3+ environments, and evaluation results demonstrate that the Combined Correlation Index achieves optimal performance with 2–3 explanation cases compared to established metrics. This tool fills an important gap in the academic software landscape by offering researchers a robust solution that enhances the scientific rigor and reproducibility of forecasting studies across disciplines.