initialize_states_using_previous_fit: Use a previous fit's initialization as initialization for a...

In timradtke/heuristika: Robust Probabilistic Forecasts to Tinker With

Use a previous fit's initialization as initialization for a new time series

Description

This function assumes that the provided y is a continuation of the time series used to derive the provided object. The first length(object$y) observations should be equal across y and object$y.

Usage

initialize_states_using_previous_fit(
  object,
  y,
  use_new_residuals_for_anomaly_candidates = FALSE
)

Arguments

object	The fitted model object returned by tulip() of class tulip
y	A time series as numeric vector, may include NAs for some (but not all) of the observations
use_new_residuals_for_anomaly_candidates	Logical (default FALSE); if TRUE, the residuals of the new time series will be used to judge whether some observations are anomaly candidates; else only the first length(object$y) observations will be used

timradtke/heuristika documentation built on April 24, 2023, 1:55 a.m.