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):
This is implemented for
print.summary.multisimsum() as well, with an additional level of nesting (by parameter).
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,
frailty2, are now bundled with
rsimsum to test the new functionality introduced above. They correspond to
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.
dfhas been renamed to
power_df, and now affects power calculations only.
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
summary.multisimsum to ease the creation of LaTeX/HTML/Markdown/reStructuredText tables.
autoplotmethods will now plot the number of non-missing point estimates/SEs by default (if the
statargument is not set by the user). The previous default was to plot bias, which might not always be available anymore since
Handling more plotting edge cases, for instance when standard errors or true values are not available;
multisimsum example in vignette on custom inputs.
Added new argument
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;
true argument of
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;
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");
Fixed bugs and errors that appeared when auto-plotting results of simulation studies with no methods being compared (#23).
autoplotsupports 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
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.
Fixed labelling bug in zipper plots (thanks to @syriop-elisa for reporting it);
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");
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.
ggplot2::label_bothfor ‘by’ factors (when included).
autoplotmethods pick the value of
multisimsumwhen 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;
pkgdownwebsite, published at https://ellessenne.github.io/rsimsum/;
autoplot method for
summary.simsum objects; when calling
summary.simsum objects, confidence intervals based on Monte Carlo standard errors will be included as well (if sensible).
Supported plot types are:
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.
rsimsumis more robust to using factor variables (e.g. as
byfactor), with ordering that will be preserved if defined in the dataset passed to
help(summary.simsum)for more details;
simsumfor testing purposes - differences are negligible, and there are some calculations in
simsumthat are wrong (already reported). Most differences can be attributed to calculations (and conversions, for comparison) on different scales.
multisimsumhas been slightly changed, with some arguments being removed and others being moved to a
controllist with several tuning parameters. Please check the updated examples for more details;
dropbigis no longer an S3 method for
dropbigis an exported function that can be used to identify rows of the input
data.framethat would be dropped by
dropbig = TRUE)are no longer included in the returned object; therefore, the S3 method
misshas been removed;
get_datais no longer an S3 method, but still requires an object of class
summary.multisimsumto be passed as input;
zipmethod has been renamed to
zipper()to avoid name collision with
ci.limitsargument (#6, @MvanSmeden). This functionality is to be considered experimental, hence feedback would be much appreciated;
relhazdataset bundled with
rsimsum 0.3.3 focuses on improving the documentation of the package.
descriptionargument to each
get_datamethod, to append a column with a description of each summary statistics exported; defaults to
rsimsum 0.3.2 is a small maintenance release:
heat()now appropriately pick a discrete X (or Y) axis scale for methods (if defined) when the method variable is numeric;
methodvarvariable to string format (if specified and not already string);
Updated code of conduct (
CONDUCT.md) and contributing guidelines (
Removed dependency on the
tidyverse package (thanks Mara Averick).
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:
lolly.multisimsumis now not required; if not provided, plots will be faceted by estimand (as well as any other
Added S3 methods for
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.
x argument to
multisimsum to include original dataset as a slot of the returned object.
miss function for obtaining basic information on missingness in simulation results.
miss has methods
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.