Digory and his Uncle Are Both in Trouble

New Features

• Added all_miss() / all_na() equivalent to all(is.na(x))

• Added any_complete() equivalent to all(complete.cases(x))

• Added any_miss() equivalent to anyNA(x)

• Added common_na_numbers and finalised common_na_strings - to provide a
  list of commonly used NA values
  #168

• Added miss_var_which, to lists the variable names with missings

• Added as_shadow_upset which gets the data into a format suitable for
  plotting as an UpSetR plot:

  airquality %>%
    as_shadow_upset() %>%
    UpSetR::upset()

• Added some imputation functions to assist with exploring missingness
  structure and visualisation:

  • impute_below Perfoms as for shadow_shift, but performs on all columns.
    This means that it imputes missing values 10% below the range of the
    data (powered by shadow_shift), to facilitate graphical exloration of
    the data. Closes #145
    There are also scoped variants that work for specific named columns:
    impute_below_at, and for columns that satisfy some predicate function:
    impute_below_if.
  • impute_mean, imputes the mean value, and scoped variants
    impute_mean_at, and impute_mean_if.

• impute_below and shadow_shift gain arguments prop_below and jitter
  to control the degree of shift, and also the extent of jitter.

• Added complete_{case/var}_{pct/prop}, which complement
  miss_{var/case}_{pct/prop}
  #150

• Added unbind_shadow and unbind_data as helpers to remove shadow columns
  from data, and data from shadows, respectively.

• Added is_shadow and are_shadow to determine if something contains a
  shadow column. simimlar to rlang::is_na and rland::are_na, is_shadow
  this returns a logical vector of length 1, and are_shadow returns a logical
  vector of length of the number of names of a data.frame. This might be
  revisited at a later point (see any_shade in add_label_shadow).

• Aesthetics now map as expected in geom_miss_point(). This means you can write
  things like geom_miss_point(aes(colour = Month)) and it works appropriately.
  Fixed by Luke Smith in Pull request
  #144, fixing
  #137.

Minor Changes

• miss_var_summary and miss_case_summary now return use order = TRUE by
  default, so cases and variables with the most missings are presented in
  descending order. Fixes #163

• Changes for Visualisation:

  • Changed the default colours used in gg_miss_case and gg_miss_var to
    lorikeet purple (from ochRe package: https://github.com/ropenscilabs/ochRe)
  • gg_miss_case
    • The y axis label is now ...
    • Default presentation is with order_cases = TRUE.
    • Gains a show_pct option to be consistent with gg_miss_var
      #153
  • gg_miss_which is rotated 90 degrees so it is easier to read variable names
  • gg_miss_fct uses a minimal theme and tilts the axis labels
    #118.

• imported is_na and are_na from rlang.

• Added common_na_strings, a list of common NA values
  #168.

• Added some detail on alternative methods for replacing with NA in the
  vignette "replacing values with NA".