Changelog

Source: NEWS.md

lighthouse 0.7.3

New functions

• ffyq() and sfyq_il() return the federal fiscal year and quarter or Illinois state fiscal year and quarter for a given date. Return format can be set using the type parameter, defaulting to numeric YYYY.Q format. These functions wrap lubridate::quarter() and complement the existing lighthouse functions ffy() and sfy_il().

New functionality

• summary_table() has new arguments .cols_group_glue and .cols_group_order to control column names and order when .cols_group_by is set. These are replacements for .cols_group_opts, which is deprecated and will be removed in a future release.

Bug fixes

• fiscal_year(), ffy(), and sfy_il() are now vectorized (fixes #23).

• summary_table() now accepts functions that do not have an na.rm or ... argument, which previously caused an error.

• strftime_no_lead():

  • now returns results with default formatting when format is unspecified (fixes #21).
  • now supports the "%OSn" conversion specification. This returns seconds with specified number of decimal places, up to 6; e.g., "%OS3" would return seconds with 3 decimals places.

• summary_report():

  • no longer issues a deprecation warning related to using !!! on a single language object (fixes #19).
  • now returns consistent column types in output tibble (fixes # 26).
  • setting .missing_label no longer throws errors in some situations (fixes #26).
  • now treats dates and datetimes as nominal by default, and will error on attempts to treat dates or datetimes as continuous or binary.

lighthouse 0.7.2

Bug fixes

• open_file(), open_location(), and in_excel() now work on MacOS (fixes #17).

Changes to row-wise aggregation functions

• Added psum() and pmean():

  • These compute “parallel” or row-wise sums or means, analogous to base::pmax() and base::pmin().

  • psum() deprecates row_sums_spss(). (psum() is a clearer and more consistent name, as its behavior is closer to that of pmin() / pmax() than rowSums().) Note that psum() has na.rm = FALSE by default whereas row_sums_spss() defaulted to na.rm = TRUE.

• Added psum_across() and pmean_across():

  • These are implementations of psum() and pmean() that take tidyselect expressions, complementing pmin_across() and pmax_across().

  • psum_across() replaces row_sums_across(), which was introduced in 0.7.0 but is now removed (closes #16).

• All p*_across() functions now accept tidyselect expressions via ... rather than cols. This makes it easier to include multiple tidyselect expressions, e.g., psum_across(var1:var9, starts_with("An")).

• Updated documentation for psum(), psum_across(), and friends. In particular, see the Details section of psum_across(), which contrasts use cases for psum() vs. psum_across().

Lifecycle changes

• coerce_na_range() is deprecated in favor of na_if_range().

Other changes

• Added a hex logo!

lighthouse 0.7.1

Bug fix

• strftime_no_lead() now removes leading zeroes only from specified components of date-times (fixes #14).

lighthouse 0.7.0

New functions

Summary functions

• summary_report() returns a summary of multiple variables, summarizing each variable based on its level of measurement.
• df_compare() is a utility for identifying differences between data frames. Given two data frames, it returns only rows and columns with differences.

Tools for missing values

• na_if_range() is a renamed, expanded, and bug-fixed version of coerce_na_range(). coerce_na_range() is retained as an alias for back compatibility.
• drop_na_rows() drops rows where all columns or a specific subset of columns are all NA.
• first_valid(), last_valid(), nth_valid() return the nth non-missing value in a vector.

Tools for character vectors

• str_c_narm() is a variant of stringr::str_c() with alternative handling of NAs.
• str_c_tidy() is a variant of stringr::str_c() that accepts tidyselect expressions.
• str_ends_any() was added to complement str_starts_any() and str_detect_any().

Tools for dates

• ffy() and sfy_il() return the federal fiscal year or Illinois state fiscal year for a given date. They wrap fiscal_year(), which returns the fiscal year based on a specified starting month.
• strftime_no_lead() formats a date without leading zeroes (e.g., “6/7/2024” instead of “06/07/2024”).
• nth_bizday() is a generalization of next_bizday().

Tools for service cascades

• cascade_fill_bwd() and cascade_fill_fwd() impute values into service cascade data based on previous or subsequent cascade steps.
• cascade_summarize() returns a summary table for service cascade data.
• These functions are not yet fully documented.

Statistical functions

• se_mean() and se_prop() compute standard error of the mean and of a proportion, respectively. se_prop() includes checks for unreliability due to low variance; see its “Details.” se_mean() replaces the ambiguously-named se(), which is now deprecated.
• ci_sig() tests if a confidence interval indicates statistical significance.
• OR_to_p1() and OR_to_p2() convert odds ratios to probabilities. They complement p_to_OR().
• dunn_test() performs Dunn’s test, a pairwise post-hoc test for following up a Kruskal-Wallis test.

Math functions

• row_sums_across() is a variant of base::rowSums() that accepts tidyselect expressions and has alternative NA handling.
• sum_if_any(), min_if_any(), and max_if_any() are variants of sum(), min(), and max() that remove NAs unless all values are NA. min_if_any() and max_if_any() were renamed from safe_min() and safe_max().

Variable transformation

• fct_collapse_alt() is a variant of forcats::fct_collapse() with options to handle non-existent values and level ordering.
• fct_na_if() is a variant of dplyr::na_if() that also removes the specified value from a factor’s levels.
• swap() swaps values between two columns. It is an unconditional variant of swap_if().

Data restructuring

• add_rows_at_value() is similar to add_blank_rows(), but allows specifying position by column values rather than row numbers. Note there have been some changes in the function interface from the pre-release version; see the “Details” section of the documentation.
• pad_vectors() pads a list of vectors with NAs to a common length.

Exporting results

• add_plot_slide() is a helper for exporting plots to PowerPoint with easier control of size and positioning.
• write_xlsx_styled() writes to .xlsx with basic column formatting.

Data visualization

• add_crossings() is a helper for creating area charts with different fills for positive vs negative values.
• after_opacity() and before_opacity() are utilities for color blending.

Other

• open_file() (alias file.open()) opens a file with its default application. open_folder() (alias dir.open()) opens a folder in the system file manager.
• Given two vectors, set_compare() returns labelled subsets of unique and shared elements.
• suppress_warnings_if() and suppress_messages_if() conditionally suppress warnings or messages based on their text.
• eq_shape() checks if two objects have the same number of dimensions and same length along each dimension.

New datasets

• gain_missing_codes is a quick reference for missing value labels used in GAIN datasets.

• state.terr.name and state.terr.abb are versions of state.name and state.abb that include US territories and the District of Columbia. state.terr.data is a data frame including names, abbreviations, and FIPS codes for US states, territories, and the District of Columbia.

Added functionality

• count_pct() and count_multiple() now support the .by argument for per-operation grouping. Integration of .by into other count_*() functions is planned for a future update.

• In summary_table(), the column of variable names can be dropped when only one variable is included by setting .var_col_name = NULL (#9).

• count_duplicates() now returns both the unique and total number of duplicated values. (e.g., c(2, 2, 4, 4) has two unique and four total values.)

• Added a missing argument to swap_if() with options for cases where the condition is missing.

• Added a warn_factor argument to try_numeric()

Bug fixes

• The .cols_group_by argument in summary_table() now produces separate columns by group (fixes #6).

• count_with_total() now produces totals for non-character columns (fixes #10).

• days_diff() now handles inputs of different types (e.g., a date and a datetime) with a warning (previously threw an error).

• Added General Election Day to holidays_il and arranged by date (fixes #1).

• Removed Inauguration Day from holidays_us.

Lifecycle changes

• rbool() has been un-deprecated. It was previously deprecated in favor of purrr::rbernoulli(), but purrr::rbernoulli() has since been deprecated itself.

• pivot_wider_alt() is defunct. Changes to tidyr::pivot_wider() made its most important functionality unnecessary. Further changes to tidyr broke it, and it was judged not worth the effort of fixing.

• na_like() and median_dbl() are deprecated. They are no longer needed given more flexible handling of mixed classes by dplyr::if_else() and dplyr::case_when() as of [dplyr v1.1.0][https://dplyr.tidyverse.org/news/index.html#vctrs-1-1-0]. (Plus na_like() was quite buggy and unreliable; resolves #2).

• row_sums_spss() deprecated in favor of row_sums_across().

• safe_min() and safe_max() renamed to min_if_any() and max_if_any(); the old names are deprecated.

• se() renamed to se_mean(); the old name is deprecated.

Other changes

• In asterisks(), changed the default for include_key from TRUE to FALSE.

• percent() and comma() re-exported from scales (#11).

lighthouse 0.6.0

New functions

• Grouping and summary functions:
  • group_with_total()
  • count_multiple()
  • count_unique()
  • count_duplicates()
  • cols_info()
• Statistical and math functions:
  • wkappa()
  • cohen_w()
  • median_dbl()
  • safe_min(), safe_max()
  • pmin_across(), pmax_across()
  • cumsum_desc()
  • scale_vec()
  • reverse_key()
• Data restructuring:
  • add_header()
  • t_tibble()
  • rev_rows()
  • fct_reorder_n()
• For working with missing values:
  • find_na_cols(), drop_na_cols()
  • n_valid(), pct_valid(), n_pct_valid()
  • discard_na()
  • null_to_na()
  • is_valid()
• For working with strings:
  • str_prefix(), str_suffix()
  • glue_chr()
• For working with dates:
  • datetimes_to_date()
  • next_bizday()
• Logical tests:
  • is_duplicate()
  • is_spss()
  • is_coercible_integer(), is_coercible_logical()
• Other:
  • gain_ss_score()

Other changes

• Added datasets for federal (holidays_us), Illinois (holidays_il), and Chestnut Health Systems (holidays_chestnut) holidays (primarily for use withnext_bizday()` function).
• Added strict argument to is_TRUE(), is_FALSE(), is_TRUE_or_NA(), and is_FALSE_or_NA()
• Improvements to set_ggplot_opts(), ggview(), and is_coercible_numeric()
• Bugfixes for in_excel(), count_na(), summary_table(), pivot_wider_alt(), print_all(), asterisks(), and coerce_na_range()
• Remove check for lighthouse updates on load

lighthouse 0.5.0

• Check if lighthouse update is available on load
• New infix operators: %all_in%, %any_in%
• Exported na_like()

lighthouse 0.4.1

• Bugfix for print_all()

lighthouse 0.4.0

• New logical tests: is_TRUE(), is_FALSE(), is_TRUE_or_NA(), is_FALSE_or_NA(), is_coercible_numeric()
• New count functions: crosstab(), count_na()
• New data transformations: scale_mad(), winsorize()
• New date functions: floor_month(), floor_week(), floor_days(), days_diff()
• Other new functions: asterisks(), print_n(), print_all(), na_to_null(), set_ggplot_opts()
• Added added optional name argument to in_excel()