NEWS

inspectdf 0.0.13

Fixed compatibility with dplyr >= 1.1.0 by replacing deprecated functions: select_if() replaced with select(where()), and mutate_if() replaced with mutate(across(where())).
Fixed critical bug in plot_cat() where bind_rows(.id = ) with unnamed lists caused failures in newer dplyr versions. Function now properly assigns column names to list elements.
Fixed issue in plot_cat() where filtering by non-existent jsd column removed all rows when plotting single dataframe summaries.
Fixed #48, inspect_num(df1, df2) with different ranges and different set of columns. Thanks to cregouby for the #51 fix.
Fixed #45, partial argument matching warning in format_size(). Changed unit = "auto" to units = "auto" in call to format(). Thanks to @salim-b for the report.
Updated CRAN checks badge URL from deprecated cranchecks.info/badges/ to new badges.cranchecks.info/ service.
Fixed ggplot2 deprecation warning by replacing size parameter with linewidth in geom_bar().

Add include_int option in inspect_cat() to allow treatment of integer columns as categorical.
Improved p-values associated with binned categorical and numeric comparisons. This is now based on a modified chi-squared test and is labelled as pval in the resulting output.
Fixed #27 ensuring plots for inspect_cat() respect any filtering or sorting of the summary output prior to show_plot(). Thanks to Roel Verbelen for the report.
Additional detail in inspect_type() comparison of two dataframes to make it easier to see which columns and types differ.

Important change: the show_plot argument has been removed from all inspect_*() functions. To generate visualisations of data frame summaries, please use the more flexible show_plot(inspect_*()) or via the pipe inspect_*() %>% show_plot().
show_plot() improvement that nudges points that might otherwise have coincided for dataframe comparisons of imbalance (for example, with inspect_imb(df1, df2) %>% show_plot())
Plots for grouped summaries: inspect_cor(), inspect_na() and inspect_num().
inspect_cor() slight speed up for dataframes with large numbers of columns.
inspect_cor() can be filtered prior to plotting, for example inspect_cor(starwars) %>% filter(abs(corr) > 0.2) %>% show_plot(). Thanks to Roel Verbelen for the suggestion
Fixed bug causing inspect_imb() to fail on certain types of factor columns. Thanks to Roel Verbelen for the report.
show_plot() has new arguments label_size, label_angle and label_color. Each provide adjustments to text annotation where applicable. Thanks to Bartosz Bursa for the suggestion.
changes to text annotation to improve how coord_flip() works on resulting plots. Thanks to Roel Verbelen for the report.

Added bytes column to inspect_mem() output, for downstream numeric comparison and consistency with inspectpd.
Added pcnt_nna column to inspect_cor() output containing the percentage of pairwise complete observations used calculated correlations. Thanks to Theo Broekman for the suggestion.
Fixed bug causing order of grouping variable in grouped inspect_ statements to be incorrect. Thanks to the report from Theo Broekman.
Removed erroneous print statement from inspect_num().

Updates to documentation throughout.
inspect_* functions now returns results by group grouped dataframes.
Added option for inspect_num() %>% show_plot() to show histograms with color palettes specified by the col_palette argument.
Fixed bug causing inspect_imb() to sometimes fail when factors present. Thanks to Doug Friedman for the report.

Fixed error causing inspect_num() to fail when columns contained all NA values. Thanks to Ryan Tanner for the report
Speed-up of inspect_cor() for large data frames with many numeric columns.
Added approximate confidence intervals and tests for method = 'kendall' and method = 'spearman' in inspect_cor().

Fix issue causing inspect_na() %>% show_plot() to fail when 0 NA present. Thanks to the report by Metin Yazici.
show_plot() now returns a ggplot2 object rather than printing the plot - thanks to Garrick Aden-Buie for the suggestion.
Dramatic speed up of inspect_cat plotting by avoiding text labels for small regions.
Added tech dataset.
Fix for text annotation of inspect_cat() plots when labels are empty strings. By default "" will be shown. Thanks to Michael Swenson for the report
inspect_cor(method = ...) argument added, thanks to suggestion from George Dontas. Options for pearson, spearman and kendall. Note that confidence intervals and tests currently only supported for pearson.
Fix error when duplicate factor labels present in inspect_cat() & inspect_imb().

text_labels autoscale size using ggfittext::geom_fit_text(). For an example see inspect_cat(). Thanks to David Wilkins for the PR.
6 different color palettes supported in show_plot() via col_palette argument. Colorblind friendly option specified via show_plot(col_palette = 1) - thanks to Richard Careaga for the suggestion.
inspect_imb().
- include_na option for categorical columns that are 100% missing, or constant are underlined in plot for easier comprehension.
inspect_cor()
- Points and whiskers changed to coloured bands for single dataframe summaries - these are easier to see when CIs are narrow.
- Points changed to bars for inspect_cor() comparison plots - makes it easier to see smaller differences in correlations.
- NA correlations omitted from inspect_cor() comparison when plotted. Ordering of correlations reversed to be consistent with returned tibble.

show_plot() function (show_plot argument in inspect_ functions will be dropped in a future version)
high_cardinality argument in show_plot() for combining unique or near-unique categories for plotting inspect_cat().
progress bars shown when processing larger datasets
Improvements to plots throughout