shadow_shift handling infinite values

[njtierney]

The handling of infinite values currently returns NaN - it should ideally return a shifted missing value and issue a warning for handling infinite values (ideally they should not be in your dataset, right?)

There should also be some way of turning off this behaviour / the loud noise that it makes

Below is some testing code for handling this.

library(naniar)
df_inf <-  data.frame(x = c(-Inf,rnorm(2), NA, Inf))

df_inf

df_inf %>%
  add_shadow_shift(x)

# use the minimum for the non infinite values
xmin <- min(df_inf$x[!is.infinite(df_inf$x)], na.rm = TRUE)

x_shift <- xmin - xmin*0.1

# set the seed here
# set.seed(seed_shift)
x_jitter <- (stats::runif(length(df_inf$x))-0.50)*x_shift*0.10

ifelse(is.na(df_inf$x),
       yes = x_shift + x_jitter,
       no = df_inf$x)

[njtierney]

Perhaps shadow_shift need to use range() instead of min() to shift missing values?