synthpop 1.9-1
---------------
CHANGES
* Removal of typos from the inference vignette.
synthpop 1.9-0
---------------
NEW FEATURES
* Two new disclosure functions calculate various disclosure measures and their
associated statistics from a set of keys combined to form a quasi-identifier.
The basic function disclosure() calculates measures for one target variable.
The other function (multi.disclosure()) calls disclosure() for a set of target
variables, defaulting to all the variables in a data set that are not in the
keys. Both functions allow numerical variables to be categorised into groups
defined by the supplied parameters. By default numeric variables are treated
as factors with levels equal to their number of distinct values.
* A new vignette on disclosure gives details and sample code for the
new disclosure functions.
* The function syn.ipf() now allows grouped margins that include all threeway
and fourway interactions, as well as twoway.
* The two syn methods that can be made differentially private (DP) (`catall` and
`ipf` can now use the Gaussian mechanism instead of the Laplace mechanism. They
each have new parameters (`noise`) to define the type of noise to add and del,
to define delta for the Gaussian mechanism. Exponential noise can also be added
but this is not recommended and a warning is printed.
* New function mergelevels() that combines levels of factors in a data.frame
using routines from the forcats package.
* New function synorig.compare() that can be used to check synthetic data created
by methods other than syn() to make sure that the data sets are comparable.
Some simple fixes can be done and the changed data sets returned.
It is called in the versions of compare.synds(), utility.gen(), utility.tab(),
utility.tables(), disclosure() and multi.disclosure() when these methods are
for synthetic data supplied as a dataframe or a list of data frames. If
fixing is possible this is done and functions continue with fixed data.
At present this only happens for first elemement of a list.
CHANGES
* Function replicated.uniques() changed. Now has additional parameter `keys`,
a vector of variable names used to define unique records. As well as checking
for replicated uniques it also checks for uniques in the synthetic data
that are present (but not necessarily unique in the original).
It now returns results as an object of class `repuniq.synds`, that can be
printed with print.repuniq.synds. Its value is a table of numbers of
different types of uniques and two vectors or matrices (if object$m>1)
that identify the records to be excluded for each type of unique
that can be excluded with the function sdc().
* Function sdc() is modified to accept input from the new replicated.uniques()
function. As well as the parameter `rm.replicated.uniques`, there is another
rm.uniques.in.orig. THis may be a more appropriate strategy for low-fidelity
synthetic data. (add ref here when I have report published).
* The functions compare.synds() and utility.tables() that calulate utility
measures comparing synthetic data to the original, now have a parameter
`print.flag` (default `TRUE`) that prints out a message as each element of the
output is calculated.
BUG FIXES
* In syn at the check for collinearity that removes some predictors, predictors
that have a passive method are not removed.
* The function utility.tables() now has an explicit parameter `max.table` that is
used by utility.tab() to override the default for the maximum table size for
any set of tables to prevent the plots from failing.
synthpop 1.8-0
---------------
NEW FEATURES
* An option in compare.synds() to add values of a selected utility measure
to names of a plot facets (`utility.to.plot` argument).
* Synthesis with "catall" and "ipf" has new features that permit the creation
of differentialy private (DP) synthetic data. Parameters are `epsilon` and
`rand`.
* New smoothing methods (`smoothing` argument): "spline" and "rmean". Now you
can also provide a single string specifying a smoothing method, which will be
then used to smooth all numeric variables in the data. You can still provide
a named list to smooth selected variables only.
* New parameters `low`, `high`, `n.breaks`, and `breaks` for utility.tables()
that allow to create a two color binned gradient and set colours for low and
high ends of the gradient.
CHANGES
* Optional parameters can be set for syn.norm(), syn.lognorm(), syn.sqrtnorm(),
syn.cubertnorm(), syn.normrank(), syn.pmm().
* syn.cubertnorm() now allows negative values.
* In multi.compare() if `msel = NULL`, pooled synthetic data copies are
compared with the original data.
* Parameter `incomplete` added to summary.fit.synds() function.
* In syn.ipf() default value of `tol` parameter changed to 1e-3 in order to
speed up convergence.
* Sublist elements of structural zero list must be named.
BUG FIXES
* In multi.compare() default colour and fill scales restored for ggplot() to
allow for more than 9 levels/synthetic data sets to be plotted.
* Automatic detection of incomplete synthesis.
* Formatting of labels for ploting numeric variables in compare.synds() that
could create duplicated values.
* Subsetting of columns in data frames that are also tibbles.
* Handling of special characters in variable names.
* Labeling of multi.compare() plots.
synthpop 1.7-0
---------------
NEW FEATURES
* New function utility.tables() summarises oneway, two- or three-way tables
of utility by utility.tab(). Plots and tables of utility measures can be
produced.
* Functions utility.gen(), utility.tab(), and compare.synds() can be used
to assess the utility of synthetic data sets that were created NOT using
synthpop. They have to be provided as a data.frame or a list.
* utility.gen() and utility.tab() have a parameter `k.syn` to indicate that
the sample size itself has been synthesised. The default value is `FALSE`
that will apply to synthetic data created by synthpop.
* The following additional statstics for each synthesis are calculated in
utility.tab(): "G" - the adjusted likelihood ratio chi-squared statistic
for comparing tables with original and synthesised values,
"JSD" - the Jensen-Shannon distance between the tables with original and
* These additional statistics are computed by utility.tab() and utility.gen():
"PO50" - the percentage over 50% of each synthetic data set where the model
used to predict real or synthetic is correct; "SPECKS" the Kolmogorov-Smirnov
distance comparing the prpensity scores for the synthetic and original data.
* Parameter `print.flag` is added to utility.gen() and utility.tab() to
suppress, if desired, output for simulations.
* compare.synds() produces a table, tab.res, that gives the oneway utility
statistics computed by utility.tab with parameter tables = "oneway"
* New parameters `plot` and `table` in compare.synds() function to indicate,
respectively, whether plots should be produced and tables printed. By default
plots are printed and tables are suppressed.
CHANGES
* Default `minnumlevels` changed from -1 to 1. This permits numeric variables
with a single distinct value as well as missing values to be synthesised
correctly.
* In utility.gen() failure to converge is changed from a warning to a failure
and message changed.
* Improved error messages in syn() to signal small samples or factor variables
with too many levels.
* Matrix of predictors x in logreg.syn() is centred and xp made to match to
improve comptutation (thanks to Heloise Gauvin for noting this problem).
* Improved error messages and help file for syn.survctree().
* In syn() variables changed to factors from character are changed back to
character in synthetic data.
* In syn() variables changed to factor from numeric because of only a few
levels, are changed back to numeric in synthetic data.
* In utility.tab() default for `digits` changed from 2 to 3.
* Several changes to a utility.gen(), including calculating p-values for
resampling methods empirically and the calculation of the percentage
correctly predicted by the model. Also changes in the saved object and in
the print method.
* Print method for utility.gen() changed to allow user to change
settings when called after the objects are created.
* In compare.synds() and multi.compare() numeric variables with fewer than
6 distinct values are changed to factors in order to make plots more
readable.
* In multi.compare() if barplot.position = "dodge", source of data
(original/sythetic) is mapped to an aesthetic `fill`.
BUG FIXES
* Problems with checks when a variable has a class attribute that is of
length > 1 corrected.
* Bug in survctree.syn() corrected.
* Bug in utility.tab() when classInt() returns duplicate values for breaks
when `style = "quantile"`, now the default, corrected.
* A synthetic version of a logical variable synthesised using function
syn.cart() is also logical.
* Passive functionality has been improved to allow checking if the passive
method would give the right answer in the original data and produce a failure
if it fails.
* Synthesis of variables with assigned labels (note that synthetic values
will not have labels).
synthpop 1.6-0
---------------
NEW FEATURES
* Incomplete synthesis is detected automatically from the details of
the methods used to synthesise a data set. Functions glm.synds(), lm.synds(),
polr.synds() and multinomial.synds() return another component `incomplete`
of their result. The function summary.fit.synds() no longer has the parameter
`incomplete`. In the previous version when a user fitted a model where some
of the variables were not synthesised and `m = 1`, summary.fit.synds() would
stop with an error. In the current version calculations are carried out
as if all variables had been synthesised (`incomplete` is set to `FALSE`)
but a warning is printed.
* New function polr.synds() to fit ordered logistic models to synthetic data
using the polr() function from the package MASS. The compare.fit.synds()
function can be used to compare the results of polr.synds() with those
based on the original data.
* New synthesising function syn.ranger() which uses a fast implementation
of random forests (contributed by Caspar van Lissa).
CHANGES
* The vignette on inference has been updated.
* synthpop website address added in a DESCRIPTION file.
* "cart" default method is explicitly defined in the syn() function.
* Correction of a Value part in the help files for syn.methods.
* Warning when numeric variables with five or fewer levels not changed
to factors.
BUG FIXES
* Method "polr" (ordered logistic regression) is not replaced by "polyreg"
(unordered logistic regression) if not necessary. A bug due to passing of an
unnecessary parameter `smoothing` removed.
* In cart.syn() when a numeric variable is synthesised and the synthetic values
of the explanatory variables cannot be classified to a final node of a tree
they are randomly assigned to one of the final children nodes.
* In glm.synds() when used with `family = gaussian` the variances are rescaled
with the residual variance estimate.
* Numeric varaiables with missing values which are not synthesised but used as
predictors for other variables are kept unchanged.
synthpop 1.5-1
---------------
BUG FIXES
* Not running an example that mostly fails due to too many small categories in
the synthesised data set.
synthpop 1.5-0
---------------
NEW FEATURES
* New methods "catall" and "ipf" that synthesise a group of variables together
at the start of the synthesis.
* New parameters `numtocat` and `catgroups` for syn(). Variables in `numtocat`
are converted into categorical variables with breaks determined by their
distribution and `catgroups` gives the target number of groups for each
variable. The data with the categorical versions are then synthesised and
finally the synthesised variables in `numtocat` are created from bootstrap
samples within the categories. This feature was developed for the second
stage of "ipf" and "catall" but can be used with any method. Variables
in `numtocat` must have a method suitable for categorical data.
* New function numtocat.syn() can be used to group numeric variables into
categories. It can be used before synthesis to create a new data frame that
can then be synthesised. It should be used instead of the `numtocat` parameter
of syn() if you want to keep the categorical variables in the synthetic data.
* The function nested.syn() can now be used with continuous variables nested
within categories. It also allows smoothing of the non-missing data.
* New function codebook.syn() that can be used to check features of data
before synthesis.
* New parameter to compare.synds() `stat` with possible values "percents" or
"counts" allows tables and plots to display counts instead of percentages in
groups.
CHANGES
* Some improvements to utility.tab():
- default for `print.tables` is to print if up to a 3-way table;
- a new parameter `useNA` to include or exclude NA values from tables;
- a new parameter `print.stats` to allow a choice of what statistics to be
printed and the default is to print only the Voas-Williamson statistic with
a simpler format.
* Format of printed output from utility.gen() has been changed to emphasise the
ratio to the expected as the most important measure.
* `mincriterion` for syn.ctree() changed to `0.9`.
* If the syn() parameter `models = TRUE`, models are stored for the variables
in the visit sequence and for missing values in the continous variables.
BUG FIXES
* utility.tab() and utility.gen() for synthetic data generated from syn.strata().
* Smoothing for "sample" method.
synthpop 1.4-4
---------------
BUG FIXES
* Compatibility with the new version of ggplot2 (2.3.0).
* Check in multi.compare() for missing argument `var`.
* Check in sdc() for data type of variables to be smoothed.
* Size of $syn when k < n and all synthesising methods without predictors.
synthpop 1.4-3
---------------
CHANGES
* In padModel.syn() the default for newly defined synthesising methods is
NOT to create dummy variables from factors before fitting models.
BUG FIXES
* Extraction of variable name from a tranformed dependant variable in
glm.synds(), lm.synds() and multinom.synds().
* In syn.polyreg() numerical variables used as predictors for the multinom()
function are scaled so as to have a range (0,1) to improve convergence. An
extra warning is included if all the preicted values are in one category,
which can result when the model fails to iterate due to sparse data.
synthpop 1.4-1
---------------
NEW FEATURES
* New version of utility.gen() and print.utility.gen(). These now implement
the methods described in Snoke et. al. and include the following:
- appropriate method of getting the null distribution is selected based on
synthesis details;
- warning messages if CART models fail to split;
- allow a seed value to be set and stored for resampling methods;
- printing Z scores from logit models using the average fit over different
syntheses.
* New function multinom.synds() to fit multinomial models to synthetic data
using the multinom() function from the package nnet.
CHANGES
* Default `cp` parameter in syn.cart() changed to 1e-8.
* Default `print.tables` parameter in utility.tab() changed to `TRUE`.
* All syn() result components that are lists, e.g. `cont.na`, have their
elements named.
BUG FIXES
* Call to classIntervals() in utility.tab() uses style = "fisher" as default
to avoid problems for variables with a small number of unique values.
* utility.tab() computes breaks for grouping from the combined data rather
than just from the observed. This avoids problems when synthetic values are
outside the range of the observed ones.
* In compare.fit.synds() the quantity real.varcov needs to be multiplied by
sigma^2 when fitting.function is "lm".
synthpop 1.4-0
---------------
NEW FEATURES
* Vignette on inference from fitted models (we are grateful to Joerg Drechsler
for his comments).
* New function syn.satcat allows saturated categorical models to be fitted
from all possible interactions of the predictor variables.
CHANGES
* utility.tab() returns p-values for utility measures rather than standardised
versions of the measures.
* `smooth.vars` parameter added to sdc() function which allows smoothing of
numeric variables in the synthesised dataset.
* If `models = TRUE` in syn(), for `logreg` and `ployreg` coefficients of the
fitted model are returned.
* Output from print.summary.fit.synds() is labelled differently according to
whether `population.inference` is TRUE or FALSE (see vignette on inference).
Also it now includes p-values and stars as are shown for lm() and glm().
* Warning messages are given in summary.fit.synds() when inference is attempted
from an invalid model where synthsis has not conditioned on any variables
that have been left unsynthesised.
* The `msel` parameter of print.summary.fit.synds() prints a table of estimates
rather than a lsiting of each fit in detail.
* Several changes in compare.fit.synds() are described in detail in the new
vignette on inference ('Inference from fitted models in synthpop'). These
include extensions of the lack-of-fit tests and confidence interval overlap
measures for different options.
BUG FIXES
* replicated.uniques() for a single variable.
* syn.cart() for logical variables without missings (thanks to bug report
by Ruben Arslan).
* syn() can be used without loading the package (thanks to reported issue
by Ruben Arslan)
* Missing data factor level <NA> as a stratum in syn.strata()
* `seed` for syn.strata().
* print.fit.synds() for syn.strata() object.
synthpop 1.3-2
---------------
CHANGES
* For consistency reasons tab.utility() changed to utility.tab() and
utility.synds() to utility.gen().
* utility.gen() under major revision and temporarily unavailable.
* Revision of utility.tab() function: a measure of fit proposed by Voas and
Williams and one proposed by Freeman and Tukey are calculated;
continous variables are categorised using classIntervals() function.
* Revision of compare.fit.synds() function: new lack-of-fit measures and mean
values for existing analysis-specific utility measures (confidence interval
overlap and absolute standardized difference between coefficient);
`return.result` parameter replaced by `print.coef` with slightly different
functionality - analysis-specific utility measures are always printed but you
can choose whether to print or not model estimates.
BUG FIXES
* compare.synds() for integer variables without missing values (thanks to bug
report by Joerg Drechsler).
synthpop 1.3-1
---------------
CHANGES
* Update of the vignette.
BUG FIXES
* compare.synds() for variables with NA values in observed but not in
synthetic data returns correct value (0) for NA category in synthetic data.
* Invalid `times` argument corrected (lists of numbers coerced to numbers).
synthpop 1.3-0
---------------
NEW FEATURES
* Storing results of CART models when `models` set to TRUE.
* Function syn.strata() for stratified synthesis.
* Function multi.compare() for multivariate comparison of synthesised and
observed data.
* Synthesising method "nested" for a variable nested within another variable.
* Tabular utility function tab.utility() for comparing contingency tables from
observed and synthesized data.
* Parameter `uniques.exclude` for the sdc() function, which can be used to
remove some variables from the identification of uniques.
* Function replicated.uniques() returns a number of unique individuals in the
original data set ($no.uniques).
CHANGES
* Synthetic values of collinear variables are derived based on the one that
is synthesised first and their method is set to "collinear". They do not
have to be removed prior to synthesis.
* Synthesising method for constant variables is set to "constant" and the
variables are not removed from the synthesised data set when
`drop.not.used = TRUE`.
* Default synthesising `method` changed to "cart".
* Default `minnumlevels` changed to -1 (during synthesis numeric variables are
not changed to factors regardless of the number of distinct values).
* Coefficient estimates and their confidence intervals are ploted in the same
order as they are presented in a tabular form.
* No message on the seed value used (it is stored in the result object).
* Formula of the model to be fitted using glm.synds() or lm.synds() can be
specified outside the function.
* Message for sdc() on number of replicated uniques also when it is equal to
zero.
* Maximum number of iterations for a multinomial model used in `polyreg` and
`polr` method increased to 1000 (`maxit` parameter). Message if the limit is
reached.
* write.syn() saves complete synds object into a file synobject_filename.RData.
* Error on exceeding `maxfaclevels` in not generated if `method` for the factor
is set to "sample" or "nested".
* For constant variables method is changed to "constant".
* Year format for variables `ymarr` and `ysepdiv` in SD2011 dataset changed
from `yy` to `yyyy`.
BUG FIXES
* Types and placement of special signs that are allowed in `rules` have been
extended and include e.g. initial and closing round bracket.
* compare.synds() provides output for logical variables.
* Synthesis of logical variables with missing values.
* Message about a change of method for a variable without predictors.
* Check for `filetype` in write.syn()
synthpop 1.2-1
---------------
BUG FIXES
* No calling var(x) on a factor x (in checks).
* No `contrasts` attribute for factors synthesised using parametric method.
* Misspelled vector name (nlevels) replaced with a correct one (nlevel).
synthpop 1.2-0
---------------
NEW FEATURES
* A new function utility.synds() for distributional comparison of synthesised
data with the original (observed) data using propensity scores.
* New measures for comparing model estimates based on synthesised and observed
data implemented in compare.fit.synds() function: standardized differences
in coefficient values(`coef.diff`) and confidence interval overlap (`ci.overlap`).
CHANGES
* No dependency on `coefplot` package.
* Default for `drop.not.used` changed to FALSE.
synthpop 1.1-1
---------------
CHANGES
* Both variable names and their column indices can be used in `visit.sequence`.
* Arguments `rules`, `rvalues`, `cont.na`, `semicont`, `smoothing`, `event`,
`denom` are specified as named lists, e.g. rules = list(marital = "age < 18")
and do not have to be specified for all variables.
* Optional arguments can be passed to synthesising functions by specifying
`funname.argname` arguments, e.g. ctree.minbucket = 5; they are
function-specific; `minbucket` removed from arguments.
* Smoothing is possible for numeric variables when synthesised with the method
"sample".
* compare() is a generic function with two methods (for class `synds` and
`fit.synds`); it replaced two separate functions.
* New argument `return.plot` for compare() method for class `fit.synds`.
* New argument `msel` for compare() method for class `synds`, which
allows comparison for pooled or selected data set(s). Results for multiple
synthetic data sets can be plotted on the same graph.
* New argument `nrow` for compare() method for class `synds`; `nrow`
and `ncol` determine number of plots per screen.
* Argument `plot.na` for compare() method for class `synds` is no longer
required and missing data categories for numeric variables are ploted
on the same plot as non-missing values.
* Argument `object` of lm.synds() and glm.synds() functions changed to `data`.
* print() method for class `fit.synds` gives by default combined coefficient
estimates only.
* summary() method for class `fit.synds` gives combined coefficient
estimates and their standard errors.
* summary() method for class `synds` with multiple synthetic data sets
provides by default summaries that are calculated by averaging summary
values for all synthetic data copies.
* Argument `obs.data` of compare.fit.synds() function changed to `data`.
* Method `surv.ctree` and `cart.bboot` changed to `survctree` and `cartbboot`.
BUG FIXES
* `denom` and `event` for variables with missing data.
* `maxfaclevels` can be increased.
* Continuous variables with missing data when zero is a non-missing value.
* Synthesis of a single variable (with or without auxiliary predictors) now
works.
synthpop 1.1-0
---------------
NEW FEATURES
* Function sdc() for statistical disclosure control of the synthesised data
set(s); function replicated.uniques() to determine which unique units in the
synthesised data set(s) replicates unique units in the original data set.
* Function read.obs() to import original data sets form external files.
* Function write.syn() to export synthetic data sets to external files and
create a text file with information about the synthesis.
* syn() has new `semicont` parameter that allows to define spike(s)
for semi-continuous variables in order to synthesise them separately.
* `lognorm`, `sqrtnorm` and `cubertnorm` methods for synthesis by linear
regression after natural logarithm, square root or cube root transformation
of a dependent variable.
* `seed` argument for syn() function.
CHANGES
* Revised output of summary.fit.synds() and compare.fit.synds();
standard errors of Z scores corrected (se(Z.syn))
(thanks to Joerg Drechsler).
* Figures for compare.fit.synds() and compare.synds() functions plotted
using ggplot2 functions.
* period.separated or alllowercase naming convention has been adopted and
parameter names `populationInference`, `visitSequence`, `predictorMatrix`,
`contNA`, `defaultMethod`, `printFlag` and `nlevelmax` have been changed to
`population.inference`, `visit.sequence`, `predictor.matrix`, `cont.na`,
`default.method`, `print.flag` and `minnumlevels` respectively.
* Default for drop.pred.only changed to FALSE.
BUG FIXES
* Rounding procedure (thanks to bug report by Joerg Drechsler).
* Warning about extra disregarded argument `family` in compare.fit.synds().