CedarKit

Github

CedarKit is a toolkit to support Convergent Cross Mapping (CCM) and Empirical Dynamic Modeling (EDM) analysis, built on top of pyEDM. It handles the parts around the actual causal-inference calculations: setting up a project, running and organizing the computations, processing the results, and visualizing them.

Projects are configured through YAML files that define variables, palettes, and run settings, so the same setup works locally or on an HPC cluster. Relationships between variables — like whether X causes Y — are represented explicitly, with a clean separation between how they’re calculated and how they’re presented.

CedarKit assembles and runs the CCM computations through pyEDM, handling details like library sizes, exclusion radius, surrogates, and lag shifting. Once results come back, it aggregates skill curves across library sizes, tracks how skill shifts between library-size bands, finds candidate peaks in skill-vs-lag curves, and picks a target lag for each relationship.

It also includes a set of plots for the standard CCM diagnostics, grid layouts for comparing across parameter sweeps, and tools for packaging results and moving them between local and HPC storage.

GPL-3.0 licensed.

PapertrailCM

(Coming soon!)