Fixed issues with nested loop plot when the simulation design is not fully-factorial (#47, thanks @mikesweeting);
Fixed wrong calculations when the same column was used in true and in by or elsewhere (#48, thanks @mikesweeting);
New options for zip plots: via the zip_ci_colors argument it is possible to customise the colour of CI lines for coverage. More details are available on GitHub (https://github.com/ellessenne/rsimsum/pull/55) and in the documentation.
Updated columns names for confidence intervals in the tt dataset;
Updated documentation regarding column names that are not allowed when calling simsum() or multisimsum(). Internals of the package have been reworked accordingly, and more informative error messages are provided to the user.
A new performance measure, relative bias, can now be calculated along with its Monte Carlo error (#41). More details on the formulae are in the introductory vignette, which has been updated accordingly.
Fixed issues with stat(level), which was deprecated in {ggplot2} 3.4.0 (#44).
Fixed an error in the calculation of the Monte Carlo standard error for relative % error in ModSE (#45, thanks @LaurenSamuels for reporting this).
Several improvements to package documentation.
This is a minor release, with the following changes:
Updated hex sticker.
Updated maintainer e-mail.
Fixed conflicts with tidy() function from the {broom} package when both {broom} and {rsimsum} are loaded at the same time. This could lead to an error of this kind:
Thanks to Theodosia Salika for reporting this.
The {pkgdown} website with the documentation has been updated to use Bootstrap 5 (and {pkgdown} ≥ 2.0.0). The new site can be found here: https://ellessenne.github.io/rsimsum/
Updated a DOI that was returning HTTP Error 503.
print.summary.simsum() now return (invisibly) a list with each section of the output, e.g. by performance measure. This is useful for printing small sections of the output, e.g. using kable() (thanks @ge-li, see discussion in #22):
library(rsimsum)
s2 <- simsum(data = relhaz, estvarname = "theta", true = -0.50, se = "se", methodvar = "model", by = c("baseline", "n"))
out <- print(summary(s2, stats = "bias"))
library(knitr)
kable(out[[1]], caption = names(out)[1], align = "r")This is implemented for print.summary.multisimsum() as well, with an additional level of nesting (by parameter).
.Rmd files), thanks to @remlapmot for reporting this (#36).power_df was passed to the control argument, it was not used (regression introduced in {rsimsum} 0.9.0). Now fixed, thanks to @Kaladani (#33).get_data() is now deprecated in favour of tidy(); get_data() still works (and is fully tested), but now throws a warning and will be fully removed some time in the future.simsum() and multisimsum() now accept multiple column inputs that identify unique methods (see e.g. #24, #30). Internally, this combines the unique values from each column factorially using the interaction() function; then, methods are analysed and reported as such. See vignette("E-custom-inputs", package = "rsimsum") for some examples.
Two new datasets, MIsim2 and frailty2, are now bundled with rsimsum to test the new functionality introduced above. They correspond to MIsim and frailty, respectively, with the only difference being that the (single) column identifying methods is now split into two distinct columns.
Improved printing for simulation studies with ‘non-standard’ way of passing true values (see e.g. #28 on GitHub);
Fixed a typo in introductory vignette;
Some internal housekeeping.
df has been renamed to power_df, and now affects power calculations only.New df argument, simsum and multisimum now accept a column in data containing a number of degrees of freedom that will be used to calculate confidence intervals for coverage (and bias-eliminated coverage) with t critical values (instead of normal-theory intervals, the default behaviour). Notably, zip plots behave accordingly when calculating and ranking confidence intervals;
Calculations for zip plots are noticeably faster now;
Added a simple kable method for objects of class simsum, summary.simsum, multisimsum, summary.multisimsum to ease the creation of LaTeX/HTML/Markdown/reStructuredText tables.
by factors from being plotted.autoplot methods will now plot the number of non-missing point estimates/SEs by default (if the stat argument is not set by the user). The previous default was to plot bias, which might not always be available anymore since rsimsum 0.8.0.Handling more plotting edge cases, for instance when standard errors or true values are not available;
Improved multisimsum example in vignette on custom inputs.
Added new argument zoom to autoplot methods: it is now possible to zoom on the top x% of a zip plot to improve readability;
Added a new example dataset from a toy simulation study assessing the robustness of the t-test. See ?"tt" for more details;
The true argument of rsimsum and multisimsum now accepts a string that identifies a column in data. This is especially useful in settings where the true value varies across replications, e.g. when it depends on characteristics of the simulated data. See vignette("E-custom-inputs", package = "rsimsum") for more details and examples;
Analogously, the ci.limits argument now accepts a vector of strings that identifies lower and upper limits for custom-defined confidence intervals from columns in data. Once again, more details are included in vignette("E-custom-inputs", package = "rsimsum");
rsimsum now correctly uses inherits(obj, "someclass") instead of class(obj) == "someclass" (#20);
Fixed bugs and errors that appeared when auto-plotting results of simulation studies with no methods being compared (#23).
autoplot supports two new visualisations: contour plots and hexbin plots, for either point estimates or standard errors. They can be obtained by selecting the argument type = "est_density", type = "se_density", type = "est_hex", or type = "se_hex".Passing the true value of an estimand (true argument) is no longer required; if true is not passed to simsum or multisimsum, bias, coverage, and mean squared error are not computed;
Passing estimated standard errors per replication (se argument) is no longer required; if so, average and median variances, model-based standard errors, relative error, coverage probability, bias-eliminated coverage probability, power are not computed.
rsimsum 0.6.1 (average and median variances were not printed).Fixed labelling bug in zipper plots (thanks to @syriop-elisa for reporting it);
Clarified that simsum and multisimsum report average (or median) estimated variances, not standard errors (thanks to Ian R. White for reporting this).
Implemented fully automated nested loop plots for simulation studies with several data-generating mechanisms: autoplot(object, type = "nlp");
Added data("nlp", package = "rsimsum"), a dataset from a simulation study with 150 data-generating. This is particularly useful to illustrate nested loop plots;
Added a new vignette on nested loop plots;
Improved ordering of vignettes.
Updated unquoting for compatibility with rlang 0.4.0;
Fixed missing details and options in the documentation of autoplot.multisimsum and autoplot.summary.multisimsum.
ggplot2::label_both for ‘by’ factors (when included).autoplot method for multisimsum and summary.multisimsum objects;simsum and multisimsum objects;autoplot methods pick the value of true passed to simsum, multisimsum when inferring the target value if stats = (thetamean, thetamedian) and target = NULL. In plain English, the true value of the estimand is picked as target value when plotting the mean (or median) of the estimated value;pkgdown website, published at https://ellessenne.github.io/rsimsum/;autoplot caused by premature slicing of by arguments, where no by arguments were included.Implemented autoplot method for simsum and summary.simsum objects; when calling autoplot on summary.simsum objects, confidence intervals based on Monte Carlo standard errors will be included as well (if sensible).
Supported plot types are:
Several options to customise the behaviour of autoplot, see ?autoplot.simsum and ?autoplot.summary.simsum for further details.
Fixed a bug in dropbig and related internal function that was returning standardised values instead of actual observed values.
rsimsum 0.4.0 is a large refactoring of rsimsum. There are several improvements and breaking changes, outlined below.
rsimsum is more robust to using factor variables (e.g. as methodvar or by factor), with ordering that will be preserved if defined in the dataset passed to simsum (or multisimsum);help(summary.simsum) for more details;simsum for testing purposes - differences are negligible, and there are some calculations in simsum that are wrong (already reported). Most differences can be attributed to calculations (and conversions, for comparison) on different scales.simsum and multisimsum has been slightly changed, with some arguments being removed and others being moved to a control list with several tuning parameters. Please check the updated examples for more details;dropbig is no longer an S3 method for simsum and multisimsum objects. Now, dropbig is an exported function that can be used to identify rows of the input data.frame that would be dropped by simsum (or multisimsum);simsum (or multisimsum) when dropbig = TRUE) are no longer included in the returned object; therefore, the S3 method miss has been removed;get_data is no longer an S3 method, but still requires an object of class simsum, summary.simsum, multisimsum, or summary.multisimsum to be passed as input;rsimsum 0.5.0.zip method has been renamed to zipper() to avoid name collision with utils::zip().ci.limits argument (#6, @MvanSmeden). This functionality is to be considered experimental, hence feedback would be much appreciated;relhaz dataset bundled with rsimsum.rsimsum 0.3.3 focuses on improving the documentation of the package.
Improvements: * Improved printing of confidence intervals for summary statistics based on Monte Carlo standard errors; * Added a description argument to each get_data method, to append a column with a description of each summary statistics exported; defaults to FALSE; * Improved documentation and introductory vignette to clarify several points (#3, @lebebr01); * Improved plotting vignette to document how to customise plots (#4, @lebebr01).
New: * Added CITATION file with references to paper in JOSS.
rsimsum 0.3.2 is a small maintenance release: * Merged pull request #1 from @mllg adapting to new version of the checkmate package; * Fixed a bug where automatic labels in bar() and forest() were not selected properly.
Bug fixes: * bar(), forest(), lolly(), heat() now appropriately pick a discrete X (or Y) axis scale for methods (if defined) when the method variable is numeric; * simsum() and multisimsum() coerce methodvar variable to string format (if specified and not already string); * fixed typos for empirical standard errors in documentation here and there.
Updated code of conduct (CONDUCT.md) and contributing guidelines (CONTRIBUTING.md).
Removed dependency on the tidyverse package (thanks Mara Averick).
Bug fixes: * pattern() now appropriately pick a discrete colour scale for methods (if defined) when the method variable is numeric.
New plots are supported: * forest(), for forest plots; * bar(), for bar plots.
Changes to existing functionality: * the par argument of lolly.multisimsum is now not required; if not provided, plots will be faceted by estimand (as well as any other by factor); * updated Visualising results from rsimsum vignette.
Added CONTRIBUTING.md and CONDUCT.md.
Internal housekeeping.
Added S3 methods for simsum and multisimsum objects to visualise results: * lolly(), for lolly plots; * zip(), for zip plots; * heat(), for heat plots; * pattern(), for scatter plots of estimates vs SEs.
Added a new vignette Visualising results from rsimsum to introduce the above-mentioned plots.
Added x argument to simsum and multisimsum to include original dataset as a slot of the returned object.
Added a miss function for obtaining basic information on missingness in simulation results. miss has methods print and get_data.
First submission to CRAN. rsimsum can handle:
Summary statistics that can be computed are: bias, empirical standard error, mean squared error, percentage gain in precision relative to a reference method, model-based standard error, coverage, bias-corrected coverage, and power.
Monte Carlo standard errors for each summary statistic can be computed as well.