In calc_shares and add_grps, change estimate argument to value. I developed this in the context of ACS data, where estimate made sense, but it's not scaling well.

Need to add a warning (e.g. "argument y is deprecated; please use z instead."), and optionally replace the old argument with the new argument. Follow advice here: http://r-pkgs.had.co.nz/release.html#compatibility

If round_digits; else if sig_digits; else

Use is_factor in case_when to decide if format given is a function, and if so, apply it to numbers.

Sometimes data is in the right order after add_grps, sometimes it isn't. Is this due to whether it's already a factor or not? calc_shares properly arranges by factor—add_grps should do this as well.

Dropping columns based on dplyr::select(-dplyr::starts_with("total")) means columns are dropped if they happen to also start with "total". Should instead use column naming that's less likely to have conflicts.

This package will be sunset soon, with functions moving into either the cwi package or the stylehaven package.

cwi

• add_grps
• bind_self
• calc_shares
• filter_down
• make_grps
• make_wide
• moe_test
• show_uniq

stylehaven

• big_number
• brk_labels
• cap_first
• clean_titles
• color_prev
• theme_din
• themed_label

Because then if the relevant column in the data frame is also called grp, there's an error.

take pipes out of functions

Add a test to make_grps to see if factor labels are character or numeric. For example:

add_grps(list(total_households = c(1, 4), cost_burden = 2, severe_cost_burden = 3), group = ratio)

works, but so should:

add_grps(list(total_households = c("owner_occupied", "renter_occupied"), cost_burden = 2, severe_cost_burden = 3), group = ratio)

Maybe by accepting a column name and values to filter for, although this doesn't seem very tidyverse-like.