Skip to contents

Create a shifted predictor

Source: R/step_epi_shift.R
step_epi_shift.Rd

step_epi_lag and step_epi_ahead create a specification of a recipe step that will add new columns of shifted data. The step_epi_lag will created a lagged predictor column, while step_epi_ahead will create a leading outcome column. Shifted data will by default include NA values where the shift was induced. These can be properly removed with step_epi_naomit(), or you may specify an alternative filler value with the default argument.

Usage

step_epi_lag(
  recipe,
  ...,
  lag,
  role = "predictor",
  prefix = "lag_",
  default = NA,
  skip = FALSE,
  id = rand_id("epi_lag")
)

step_epi_ahead(
  recipe,
  ...,
  ahead,
  role = "outcome",
  prefix = "ahead_",
  default = NA,
  skip = FALSE,
  id = rand_id("epi_ahead")
)

Arguments

recipe

A recipe object. The step will be added to the sequence of operations for this recipe.

...

One or more selector functions to choose variables for this step. See recipes::selections() for more details.

lag, ahead

A vector of integers. Each specified column will be the lag or lead for each value in the vector. Lag integers must be nonnegative, while ahead integers must be positive.

role

For model terms created by this step, what analysis role should they be assigned? lag is default a predictor while ahead is an outcome.

prefix

A character string that will be prefixed to the new column.

default

Determines what fills empty rows left by leading/lagging (defaults to NA).

skip

A logical. Should the step be skipped when the recipe is baked by bake()? While all operations are baked when prep() is run, some operations may not be able to be conducted on new data (e.g. processing the outcome variable(s)). Care should be taken when using skip = TRUE as it may affect the computations for subsequent operations.

id

A unique identifier for the step

Value

An updated version of recipe with the new step added to the sequence of any existing operations.

Details

The step assumes that the data's time_value column is already in the proper sequential order for shifting.

Our lag/ahead functions respect the geo_value and other_keys of the epi_df, and allow for discontiguous time_values. Both of these features are noticably lacking from recipe::step_lag(). Our lag/ahead functions also appropriately adjust the amount of data to avoid accidentally dropping recent predictors from the test data.

The prefix and id arguments are unchangeable to ensure that the code runs properly and to avoid inconsistency with naming. For step_epi_ahead, they are always set to "ahead_" and "epi_ahead" respectively, while for step_epi_lag, they are set to "lag_" and "epi_lag, respectively.

See also

Other row operation steps: step_adjust_latency(), step_growth_rate(), step_lag_difference()

Other row operation steps: step_adjust_latency(), step_growth_rate(), step_lag_difference()

Examples

r <- epi_recipe(covid_case_death_rates) %>%
  step_epi_ahead(death_rate, ahead = 7) %>%
  step_epi_lag(death_rate, lag = c(0, 7, 14))
r
#> 
#> ── Epi Recipe ──────────────────────────────────────────────────────────────────
#> 
#> ── Inputs 
#> Number of variables by role
#> raw:        2
#> geo_value:  1
#> time_value: 1
#> 
#> ── Operations 
#> 1. Leading: death_rate by 7
#> 2. Lagging: death_rate by 0, 7, 14