Phase 2: statscore-timeseries — ARIMA, smoothing, stationarity + Python

[armanrasta]

Summary

Time series analysis: stationarity tests, smoothing, ARIMA, decomposition. Python bindings ship alongside.

Rust (statscore-timeseries)

• ADF, KPSS stationarity tests
• EMA, Holt, Holt-Winters smoothing
• AR, MA, ARMA, ARIMA
• Trend + seasonal + residual decomposition

Python bindings (statscore.timeseries)

• fit() / forecast() API
• AIC/BIC match R forecast package

Done when

• ARIMA AIC/BIC validated vs R
• Python forecast API works end-to-end