smooth.SMA.rmultistep

SMA.rmultistep(h=10)

Return the (T-h) × h matrix of rolling in-sample multistep forecast errors.

For each origin t, computes the h-step-ahead forecast and the corresponding errors against the realised observations. Must be called after fit().

Parameters:

h (int) – Forecast horizon (number of steps ahead). Default 10.

Returns:

Shape (T-h, h) where T is obs_in_sample.

Return type:

DataFrame

---

Parent Class: SMA