finalfit 1.0.5

formula interface added to finalfit() and summary_factorlist().
weights argument added to summary_factorlist() for weighted tables.
finalfit() updated and arguments to underlying models can now be passed directly.

finalfit 1.0.4

missing_plot() bugs fixed, many thanks @nathansam. #72
finalfit() for CPH models now provides column proportions by default, many thanks corneliushennch. #74
summary_factorlist_stratified(): beta testing for stratified tables.
rm_empty_block() added: remove rows where all specified variables are missing.

add_row_total in summary_factorlist() now can include proportion of complete data via include_row_totals_percent argument.
Robust standard error approach added to vignette.
Support added for weights etc. to be included in coxphmulti().
“Univariable” mixed effects models including random effects added to vignette.
summary_factorlist(): non-parametric continuous variables now defaults to Q1 - Q3 rather than single figure IQR.
ff_interaction(): default factor separator changed from “|” to “_” and variable separator from “__” to “_” given incompatibilities with packages such as brms.
Bug to coefficient_plot() fixed to bring back point estimates.
na_to_prop = FALSE in summary_factorlist() to not include missing data in column proportions of categorical data.

ff_relabel_df() added to allow passing data frame / tibble with labels directly at bottom of pipe.

ff_relabel() tightened to allow mismatch between available data and labels.
missing_compare() code updated to allow arguments to be passed to new summary_factorlist().
Quadratic terms (I(var1^2) etc.) are now better supported in finalfit().
cluster(), frailty() and strata() terms shown in finalfit() regression tables as an indicator they have been included in model.
or_plot() remove_ref bug fix.
ff_newdata() bug fix.
Minor changes to accommodate dplyr 1.0.0.
Plots bug fix when using remove_ref argument.

First official release.
Re-write of major functions to remove older dependencies, improve performance and add functionality.
summary_factorlist() completely rewritten. New column and row summary functions. Alternative statistical tests included. Finer control over continuous variable behaviours.

Built on R 3.6.1.
Better support for multiple imputation. fit2df() function for mipo objects. See missing data vignette/article for examples.
Easier now to build on a finalfit() table by including keep_fit_id = TRUE. See ff_merge() documentation for details.
Issues with update of tidyr::spread() in summary_factorlist() so updated to tidyr::pivot_wider().
ff_column_totals() added to be used in combination with summary_factorlist().
ff_row_totals() added to be used in combination with summary_factorlist().
ff_percent_only() added to be used in combination with summary_factorlist(). #25
ff_remove_p() can be applied to any condensed finalfit output to remove the p-value. #26
finalfit() now takes column = FALSE to provide row proportions. #26
check_recode() added.
remove_labels() now works for tibbles. #28
summary_factorlist() includes argument cont_range = TRUE to include quartiles Q1 and Q3 when median for continuous variables. #29

data(wcgs) added.
summary_factorlist() geometric sd added.
AIC added to lm metrics.
ff_label() now does not add class “labelled”.
Removed capacity for glmmulti() and lmmulti() to run multiple models from multiple dependent variables. It wasn’t used and the list generated was inconvenient for passing output to other functions such as ggfortify::autoplot().
ff_permute() re-written to allow many more options for producing intermediate models.

coxphuni() and coxphmulti() now take the other library(survival) functions survival::strata() and survival::cluster().
Fixed hr_plot() axis title edit option.
Add option to remove reference level (remove_ref = TRUE) to or_plot(), hr_plot() and coefficient_plot().
summary_factorlist() digit rounding option added.
summary_factorlist() geometric mean option added.
Level label removed for continuous variables from or_plot(), hr_plot() and coeffient_plot().

Bug in or_plot() and hr_plot() introduced in 0.9.2 because of new total column specification.

Competing risks time-to-event regression now supported via cmprsk::crr(): crruni(), crrmulti() and fit2df().
Complex stratified sampling now supported via library(survey): svyglmuni(), svyglmmulti() provide support for. #13
summary_factorlist() total column now summarises continuous variables. #17 #21
summary_factorlist() can now take any Hmisc:::summary.formula argument, such as catTest = catTestfisher.
catTestfisher() added.
finalfit_permute() added.

glmuni(), glmmulti(), lmuni(), lmmulti() now all take weights and any other glm() or lm() argument. #13
summary_factorlist() rework. Now supports any number of factor levels in dependent. #14 #15
summary_factorlist() now provides total count for continuous variable. #17

or_plot() bug fix
ff_remove_ref() added. #12
glmmixed() and lmmixed() now support random gradient models, and all complex lme4 specifications.
Data preparation vignette added.

ff_glimpse() re-written to remove psych dependency
missing_glimpse() added: single data frame describing all variables and missing values
ff_interaction() added: create variable for an interaction between two factors
ff_label() added: easily add label to variable in dataframe
ff_newdata() modified to take dataframe without requirement for dependent and explanatory arguments
summary_factorlist() modified to allow user to change number of unique factor levels at which a variable a continuous variable is converted to a factor (cont_cut). #9
fit2df() and its internal function extract_fit modified to take confint_type and confint_level.

lmuni(), lmmulti(), lmmixed(), glmuni(), glmmulti(), glmmixed(), coxphuni(), coxphmulti()

metrics_hoslem() is the first of a number of ‘metrics’ functions which will be introduced.