panel_diff {panelWranglR}

Tidy panel differencing

Description

Efficient, tidy panel differencing

Usage

panel_diff(data, cross.section, time.variable = NULL, diff.order = 1,
  lags = 1, variables.selected = NULL, keep.original = FALSE)

Arguments

data

The data input, anything coercible to a data.table.

cross.section

The cross section argument, see examples.

time.variable

The variable to indicate time in your panel. Defaults to NULL, though it is recommended to have a time variable.

diff.order

The number of applications of the difference operator to use in panel differencing. Defaults to 1.

lags

The number of lags to use for differences. Defaults to 1.

variables.selected

A variable selection for variables to difference, defaults to NULL and differences ALL variables.

keep.original

Whether to keep the original undifferenced data, defaults to FALSE.

Details

Works on a full data.table backend for maximum speed wherever possible.

Value

The differenced data.table which contains either only the differenced variables, or also the original variables.

Examples

X <- matrix(rnorm(4000),800,5)
tim <- seq(1:400)
geo_AT <- rep(c("AT"), length = 400)
geo_NO <- rep(c("NO"), length = 400)
both_vec_1 <- cbind(tim,geo_NO)
both_vec_2 <- cbind(tim,geo_AT)
both <- rbind(both_vec_1,both_vec_2)
names(both[,"geo_NO"]) <- "geo"
X <- cbind(both,X)

panel_diff(data = X,
           cross.section = "geo_NO",
           time.variable = "tim",
           diff.order = 1,
           lags = 1,
           variables.selected = c("V3","V4"),
           keep.original = TRUE)