Forecast from a segment of a segmentation

The point of detecting a regime change is usually that the earlier regime should no longer inform what happens next. predict() on a ggcpt forecasts from one segment only (by default the last), with the segment's own model.

Usage

# S3 method for class 'ggcpt'
predict(
  object,
  newdata = NULL,
  h = 1,
  segment = "last",
  models = NULL,
  level = 0.95,
  ...
)

Arguments

object

A ggcpt object.

newdata

Optional data frame of covariates (for a formula fit, or a segment model with covariates), or of time positions.

h

Forecast horizon, when newdata is not given: the positions n + 1 to n + h. Defaults to 1.

segment

Which segment to forecast from: "last" (the default) or a segment number.

models

Optional cpt_segment_models() result; built with its defaults when omitted.

level

Level of the prediction interval. Defaults to 0.95.

...

Further arguments for the model's predict() method.

Value

A tibble with time (when forecasting by horizon), .pred, and .lower/.upper when the model supplies a prediction interval.

See also

Other segment models: cpt_segment_models()

Examples

set.seed(1)
fit <- cpt_detect(c(rnorm(100), rnorm(100, 3)), method = "pelt")
predict(fit, h = 3)
#> # A tibble: 3 × 4
#>    time .pred .lower .upper
#>   <int> <dbl>  <dbl>  <dbl>
#> 1   201  2.96   1.05   4.87
#> 2   202  2.96   1.05   4.87
#> 3   203  2.96   1.05   4.87