ycorr

ycorr is a software library and toolkit for computing correlations and cross-correlations in data analysis, with a focus on time series. It implements multiple correlation metrics, including Pearson, Spearman, and Kendall tau, as well as partial correlation and robust estimators that resist outliers. The library supports both static datasets and streaming inputs, enabling rolling-window and expanding-window analyses to track how relationships change over time. Cross-correlation features allow examination of lagged relationships with configurable maximum lag and significance testing options.

Implementation and API: The core engine is written in Rust for performance, with bindings to Python and other languages to ease integration. The API is designed to be consistent across platforms and data representations. Users can provide data as arrays or labeled data frames; outputs include a correlation matrix, optional p-values, and confidence intervals. The package includes controls for missing data handling, normalization, and outlier-robust options, as well as parallel computation to accelerate large workloads.

Applications and usage: ycorr is used across finance, neuroscience, and environmental science to quantify relationships between variables and to explore temporal dependencies. It supports large-scale analyses, reproducible workflows, and integration with common data-science stacks.

License and development: The project is open-source and community-maintained, with public documentation, tutorials, and example notebooks. Ongoing development focuses on expanding metric support, improving nonparametric and nonlinear measures, and enhancing interoperability with other statistical tools.