- fix infinite loop in matchImpute in case all observations of a variable are missing
- remove parameter metric from kNN because it was not used
- add function xgboostImpute for using a simple xgboostModel to impute
- default robust regression method for irmi for numeric variables changes from rlm to lmrob.
- ordFun as parameter of kNN to control the function applied to ordinal variables
- methodStand option in gowerD and kNN to switch between range and interquartile range for the standardization of numerical variables
- donorcond in kNN and hotdeck extended so it also accepts NULL as list element and multiple conditions as character vector
- error message in
hotdeck()
when ord_var and variable overlap - family argument of class 'family' now work in regressionImp
- add new vignettes explaining all remaining imputation methods (
irmi()
,kNN()
,hotdeck()
andregressionImp()
). Thanks @wolfgangrannetbauer (#44, #45) - Allow missing and imputed values in several visualization functions.
- The new functionalities are showcased in the new visualization vignette (#46). Thanks @wolfgangrannetbauer!
- Add
tableMiss()
: A table that highlights missing and imputed values via colors (#47). - Bug fix for kNN (#48) Thanks @torockel
- extend documentation with new vignettes and pkgdown
- add rangerImpute() to impute values with
ranger::ranger()
(#35) - remove support for survey objects (#36)
- remove exports for VIMGUI (#40)
- change data.table dependency from depends to imports (#41)
- bugfixes for
irmi()
with logical and integer columns (#42)
- updates for
gowerD()
- separate help pages for
maxCat()
andsampleCat()
- remove links to certain packages
- bcancer
- brittleness
- colic
- diabetes
- food
- pulplignin
- toydataMiss
- wine
- data set
collision
added
- fixed a bug in the distance computation when different variable types are used
- added
imp_var
andimp_suffix
toirmi()
, so it is more consistent with the other functions (#27)
- added parameter addRF, only RF to
kNN()
, to use random forest in combination withkNN()
- testthat package used for all tests
- added travis automatically building/checking
- covr for code coverage (vis functions are currently not covered by any tests)
- new imputation function
matchImpute()
for imputing randomly within groups
- remove handling of
impNA
inhotdeck()
- add regression tests
- bugfix for
irmi()
with factors (#13). Thanks @Deleetdk - bugfix with colorspace package (#4)
- use ordered logistic regression for ordinal variables in
irmi()
(#23) - add support for ordered factors (#7)
- bugfixes (#8, #9)
- bugfixes for
kNN()
anddata.table
- bugfix for labelled vars in
hotdeck()
- add JSS citation
- bugfix, if a
data.frame
is passed toirmi()
(#6)
- new option for
kNN()
:weightDist
to use the distances for the k nearest neighbours as weights - The R function
which.minN()
is not used anymore, instead there is a C++ function,kNN()
is now about 1.6 times faster on a replication (100x) of the sleep dataset - Bytecompile is enabled
- bugfix wrong observations marked as imputed in
hotdeck()
- random sorting is now used in
hotdeck()
if noord_var
is defined
- bugfix for
hotdeck()
withmakeNA
- bugfix for the computation of distances for ordered variables
- new option for
kNN()
useImputedDist
if the imputed values of a variable should be used in subsequent imputation of another variable.
- bug fixed in
irmi()
with newer version of nnet (multinom) and if residual scale can not be computed (noise) - Improvement Gower dist with only missing values in data.x or data.y
- new parameter
modelFormula
inirmi()
- bug fixes in
irmi()
- updated
hotdeck()
based on data.table -> faster and quite stable - bug fix if range of a variable is 0 in gower.dist
- small fixed
kNN()
- small bugfix for using
makeNA
inkNN()
- "Nothing to impute"-Error is now a warning
imp_var
now updates existing TFimp_vars
(with warning)
- new pacakge VIMGUI contains all GUI functions
- vignettes moved to VIMGUI
- new imputation function
regressionImp()
- roxygen style comments -> help files