statsforecast plugins

by Jeremy Longshore

Run Nixtla-style baseline forecasting models (AutoETS, AutoTheta, SeasonalNaive) on M4 / custom CSV datasets from inside Claude Code. Pure statsforecast — no API keys.

by Jeremy Longshore

Run Nixtla statsforecast / TimeGPT forecasting on BigQuery time-series data with SQL-injection-validated identifiers and exponential-backoff retry on transient GCP errors. Pull from BigQuery, forecast, write results back.

by Jeremy Longshore

Real migration assistant for moving from Prophet, statsmodels, or sklearn forecasting to Nixtla TimeGPT or StatsForecast. AST-based code analysis, pandas-based data transformation to canonical (unique_id, ds, y) schema, code generation across 5 source/target combinations, real side-by-side accuracy comparison, and phased migration plans.

by Jeremy Longshore

Automated content discovery for Nixtla / time-series practitioners. Searches the web + GitHub for forecasting content, generates AI summaries, and publishes formatted digests to Slack on a schedule.

by Jeremy Longshore

Head-to-head benchmark of TimeGPT vs open-source StatsForecast on user data. Real model fits (AutoETS, AutoTheta, SeasonalNaive + optional NixtlaClient), real metrics (sMAPE, MASE, MAE, RMSE), wall-time comparison, rule-based recommendations.