NEWS
party 1.3-17 (2024-08-17)
party 1.3-16 (2024-07-18)
party 1.3-15 (2024-04-29)
party 1.3-14 (2023-11-28)
- fix print statement in RandomForest.c
party 1.3-13 (2023-03-17)
- do not create variables in workspace
party 1.3-12 (2023-02-28)
party 1.3-11 (2022-10-07)
party 1.3-10 (2022-04-26)
- proper use of inherits in two instances
party 1.3-9 (2021-09-27)
- #define USE_FC_LEN_T as requested by CRAN 2021-09-25
party 1.3-8 (2021-08-23)
party 1.3-7 (2021-03-03)
party 1.3-6 (2021-02-08)
- fix LaTeX problem
- documentation updates
party 1.3-5 (2020-06-12)
- allow varperm to change in calls to R_get_nodeID (feature request by Dries Debeer)
party 1.3-4 (2020-03-05)
party 1.3-3 (2019-03-06)
party 1.3-2 (2019-03-01)
- move ExpectCovar and ExpectCovarInfluence classes from coin to
party
party 1.3-1 (2018-08-08)
- varimpAUC() now calls varImp::varImpAUC()
party 1.3-0 (2018-04-20)
- New scale argument for predict(cforest, scale = TRUE, ...).
For simple regression forests, predicting
the conditional mean by nearest neighbor weights with
scale = TRUE is now
equivalent to the aggregation of means. The unscaled version
proposed in <doi:10.1002/sim.1593> can be obtained with
scale = FALSE.}
party 1.2-4 (2017-12-13)
- Bonferroni sometimes overcorrected (by taking constant variables into account).
Spotted by Heidi Seibold.
- Bugfix for long-standing problem in conditional variable importances
by Dries Debeer.
party 1.2-3 (2017-04-12)
- predict(cforest, OOB = TRUE) in party < 1.2-0 was _correct_
and version 1.2-0 to 1.2-2 contained in incorrect "correction".
The test set error and the out-of-bag error are now (again)
roughly equal. There won't be any correlation between
observations and out-of-bag predictions when the response is
independent to all explanatory variables. Out-of-bag predictions
in package partykit (>= 2.0-0) are now based on the same
out-of-bag weights and the results from both packages are
roughly equivalent. The issue was first spotted by
Julian Karch and later by Andrew Sage.
- comment on false alarms by maacheck / bcheck (email by
Tomas Kalibera April 10, 2017) in party/src/*c
party 1.2-2 (2017-02-27)
- free memory in src/Distributions.c (pointed out by
Martin Binder)
- register C routines
party 1.2-1 (2017-02-09)
- fix a memory leak occuring while searching for surrogate
splits in the process of predicting the terminal node ID
- surrogate splits might have used statistics with slightly
incorrect weights
- party::ctree gives _same_ results as partykit::ctree
(version >= 2.0-1 from R-forge)
party 1.2-0 (2017-02-06)
- predict(cforest, OOB = TRUE) relied on incorrect weights
and gave suboptimal predictions. varimp() was not affected.
The effect was pointed out by Julian Karch when comparing
party::cforest to partykit::cforest
party 1.1-2 (2016-11-28)
- remove ctree_memory(), also from examples
party 1.1-1 (2016-11-23)
party 1.1-0 (2016-11-21)
- don't change values of R objects in C code, this caused
problems with byte compiling. ctree_memory() had to be
removed in this process.
party 1.0-26
- compare with libcoin instead of coin
party 1.0-25 (2015-11-05)
party 1.0-24
- initialize, show needed import
party 1.0-23 (2015-08-16)
- update Rout.saves to coin 1.1-0: logrank_trafo now
has the same sign for each tie handling method which
changes some of the cut-points
party 1.0-22 (2015-07-28)
party 1.0-21 (2015-06-06)
- reference output updated for recent versions of R
party 1.0-20 (2015-02-18)
- is.na.Surv is now an S3 generic in survival 2.37-7
party 1.0-19 (2014-12-18)
- directly link to mvtnorm and thus make duplication of
FORTRAN code obsolate.
- export prettytree (for package mlmeta by Andrew Z)
party 1.0-18 (2014-11-10)
- internal call to svd acummulated memory until the
garbage collector was able to clean up (only after the
whole forest was grown). Fix by Philip Montgomery
<[email protected]>.
party 1.0-17 (2014-09-17)
- move initialize methods of coin classes to coin and
re-import them
party 1.0-16 (2014-08-19)
party 1.0-15 (2014-06-30)
- allow for missing predictor values in new varimp for
right-censored responses
- introduce varperm argument to R_get_nodeID
party 1.0-14 (2014-05-30)
- add experimental varimp() for right-censored responses
party 1.0-13 (2014-02-01)
- deal with protection issues in R-devel
party 1.0-12 (2014-01-17)
party 1.0-11 (2013-12-14)
- deal with empty levels for teststat = "quad"
(bug reported by Wei-Yin Loh <[email protected]>)
party 1.0-10 (2013-10-24)
- new demo "strucchange-perm"
- proximity has newdata argument
party 1.0-9 (2013-09-02)
- export initVariableFrame (for usage in RWeka only)
- fix NAMESPACE problems
- data is in TH.data, not ipred anymore
party 1.0-8 (2013-07-11)
- update mvtdist
- documentation update
party 1.0-7 (2013-06-05)
- ctree and cforest ignored the scores argument for the
response. Spotted and fixed by Silke Janitza
<[email protected]>.
party 1.0-6 (2013-01-09)
- use hist()$density instead of hist()$Intensities as
reported by R 3.0.0 to be
party 1.0-5 (2013-01-07)
- fix class problem reported by R 3.0.0 to be
party 1.0-4 (2012-12-19)
- export varimpAUC()
- add where() method for RandomForest class
party 1.0-3 (2012-09-26)
- new function varimpAUC() for AUC-based variable importance measures
- the ‘weights’ argument in cforest() can now be used to define
alternative case weights for each of the trees in the forest.
party 1.0-2 (2012-04-26)
- version 1.0-0 introduced a new bug in plotting Kaplan-Meier curves
when fixing the weight problem in survfit; the x-axis labels where
incorrect. Spotted by Sarah Vowler <[email protected]>.
party 1.0-1 (2012-03-23)
- fix -Werror warnings suggested by Murray Stokely <[email protected]>
- make sure no factor variables are evaluated as surrogate split
- fix confusing warning messages in varimp and check
for non-unity weights
party 1.0-0 (2012-02-28)
- varimp can now partially deal with missing values
- Kaplan-Meier curves are now computed _after_ removing
zero case weights (see comment in ?survfit); plots may change
slightly.
- use cforest_unbiased instead of cforest_control/cforest_classical
in examples and regression tests
party 0.9-99996 (2011-12-20)
- don't ship doxygen documentation with package
- ctree() and cforest() don't handle ‘-’ in formulae,
warning added to Rd files
party 0.9-99995 (2011-11-02)
- observations with missing values in newdata might not have gone
with the majority in predict(ctree, newdata) when no surrogate
split was applicable. Spotted and fixed by
Vik <[email protected]>.
party 0.9-99994 (2011-08-24)
- reuse classes exported by coin
party 0.9-99993 (2011-07-29)
- attach NAMESPACE properly in tests
party 0.9-99992 (2011-05-02)
- fix partial matching problems
party 0.9-99991 (2010-11-19)
party 0.9-9999 (2010-09-13)
- add update slots to tree and forest objects (later to be used
for crossvalidation)
party 0.9-9998 (2010-08-23)
- allow for internal permutations in predict via R_predict2
party 0.9-9997 (2010-07-06)
- update mvt.f (see mvtnorm)
party 0.9-9996 (2010-06-25)
- print progress bar while forest grows (via trace = TRUE
in cforest_control())
party 0.9-9995 (2010-05-19)
- assignment of observations based on surrogate splits
during fitting might have failed; observations went
with majority (in a sub-optimal way)
party 0.9-9994 (2010-04-15)
- varimp() gives warning for non-unity weights
- "no admissible split found" warnings disabled
party 0.9-9993 (2010-03-05)
- just internals needed for mboost
party 0.9-9992 (2010-02-26)
- Bug fix in mob plotting for anova models
party 0.9-9991 (2010-01-27)
- Bug fix in error handling of mob().
party 0.9-999 (2009-09-24)
- minbucket was erronously interpreted as minbucket + 1
- p-values for teststat = "max" were wrong
in the uncommon case of zero variances
party 0.9-998 (2009-06-17)
- mtry may vary with tree depth
- finding surrogate variables may have failed, fixed.
party 0.9-997 (2009-06-04)
party 0.9-996 (2009-03-27)
party 0.9-994 (2009-02-11)
- fix bug in new varimp function (which was triggered
by splits in categorical variables)
party 0.9-993 (2009-01-29)
- new C-level function R_remove_weights
party 0.9-992 (2009-01-19)
party 0.9-991 (2009-01-14)
- new conditional variable importance measure
(Strobl et al., 2008, BMC Bioinformatics) available.
party 0.9-99 (2008-10-28)
- a suboptimal split was implemented when the best
split corresponded to the last row of data due to
a coding error. This should have happend only very
rarely, however, the glaucoma example is affected.
party 0.9-98 (2008-10-15)
- don't compute ordering of censored variables (thanks to
Brian D. Ripley for the report)
- experimental version of partialPlot for cforests
- teststat = "quad" and testtype = "Univariate" produced
wrong criteria in the presence of constant exploratory variables.
Spotted by Sebastian Wietzke <[email protected]>
party 0.9-97 (2008-06-27)
- predictions for new data containing missing values were
wrong. Bug spotted by Wei-Yin Loh <[email protected]>
- vignette updates
party 0.9-96 (2008-06-08)
- make searching for surrogate splits a little more stable
party 0.9-95 (2008-02-28)
- proximity() now gives _relative_ numbers
(similar to ‘randomForest’), not absolute numbers
long-standing feature request by Henric Nilsson
party 0.9-94 (2008-01-30)
party 0.9-93 (2007-11-30)
- It is now an error to ask for more surrogate variables
than actually available. Thanks to Henric Nilsson.
party 0.9-92 (2007-11-22)
- predict(<ctree-object>) computed predictions
for observations with zero weights (by passing them down the tree).
This was already done this way by predict(<cforest-object>) but resulted
in an error for ctree()s.
- minbucket added to ctree_control()
party 0.9-91 (2007-09-27)
- bugfix: NAs in factors lead to segfaults on 64bit machines
as reported by Carolin Strobl <[email protected]>
- related to this: be more careful when splitting in
variables containing missing values, especially factors.
- add examples to regression tests
- new function ‘proximity’ (feature request by Henric Nilsson)
party 0.9-90 (2007-09-24)
- new defaults for ‘control’ argument of ‘cforest’: unbiased
forests are now grown by default
- new defaults for binary classification tree visualization:
stacked bar charts are used by default.
party 0.9-14 (2007-07-23)
- fix problem in ‘varimp’ introduced by new weights handling
party 0.9-13 (2007-07-23)
- update ‘mvt.f’
- improve the memory footprint of RandomForest objects
substancially (by removing the weights slots from each node).
More careful revision of the predict and fitting code is needed.
- predictions for cforest with multivariate response are
now returned in a matrix (should be a data.frame)
- remove confusing ‘no admissible split found’ warning
(which simply meant that the ‘minsplit’ wasn't met).
party 0.9-12 (2007-06-28)
party 0.9-11 (2007-06-21)
- populate ‘where’ slot in ‘mob’ objects.
- ‘mob_control’ now has an ‘na.action’ argument
- cforest with replace = FALSE, fraction = ...
can now deal with weights (spotted by Carolin Strobl
<[email protected]>)
- interface to LAPACK (‘dgesdd’ for computing a SVD)
directly. As a side-effect, the svdmem class can be
simplified and no memory allocation during
tree growing is necessary (which results in a measureable
performance gain)
party 0.9-10 (2007-02-28)
- internal infrastructure for fit(x, y) interfaces,
including a VariableFrame method for matrices.
- deal with factors with only one level and variables where
all observations are NA
- don't compute design matrices in ‘ModelEnvFormula’
party 0.9-9 (2007-02-02)
- internal changes to data objects: joint transformations
only necessary for response variables (new class ‘ResponseFrame’)
but we need to distinguish between testing (using ranks, for
example) and prediction (means of raw data!).
Problem spotted by Henric Nilsson <[email protected]>.
party 0.9-8 (2007-01-22)
- thicker median lines in boxplots
party 0.9-7 (2007-01-15)
- fix problems with multivariate response variables
(spotted by Henric Nilsson <[email protected]>)
party 0.9-6 (2006-11-28)
- R-2.5.0 does not allow to change the storage.mode of factors
party 0.9-5 (2006-10-27)
- adjusted p values computed with testtype = "MonteCarlo"
did not take dimensionality of the underlying linear statistics
into account. It is now ensured that _p values_ are resampled,
not test statistics.
- fix escape problems
party 0.9-4 (2006-09-15)
- make sure ‘maxsurrogates’ exceeds number of input variables
- fix problems reported by valgrind
party 0.9-3 (2006-09-08)
- don't call RNG functions in mvt (this is done in C code, once)
party 0.9-2 (2006-09-07)
- mob() can't currently handle missing values,
they are removed and a warning is given.
- add ‘methods’ and ‘vcd’ to Depends:
party 0.9-0 (2006-08-31)
- node_barplot() can now also draw stacked barplots.
- simplify handling of ordinal variables. It is now possible
to have multiple (ordinal) response variables.
- some documentation updates
- fix problems with NA handling for factors
- make sure minsplit criterion applies in the presence of missing
values. Bug spotted by Han Lee <[email protected]>.
party 0.8-8 (2006-08-23)
- export ‘varimp’ (variable importance measures for cforest())
- documentation updates
- add codetools checks
party 0.8-7 (2006-07-18)
- REAL(NULL) now gives an error in C code and scores could be
potentially NULL: fixed for the time being.
- add ‘show’ method for BinaryTree and RandomForest objects
party 0.8-6 (2006-05-22)
- add some validity checks for hyper parameters
- add function ‘varimp’ for computing variable importance
measures for some random forests (this is still experimental and
thus not exported yet, so use ‘party:::varimp’)
party 0.8-5 (2006-04-24)
- all hyper parameters to ‘cforest’ are now controlled using
objects of a new class ‘ForestControl’ which can be created using
‘cforest_control()’.
- sampling of a fraction of the observations without replacement
is now available for the ‘cforest’ implementation of random forest.
party 0.8-4 (2006-04-12)
- renamed argument ‘teststattype’ to ‘teststat’ with defaults
c("quad", "max") in order to be compatible with the ‘coin’ package,
version >= 0.4-5
- use ‘1 - (1 - p)^k’ instead of ‘p * k’ for Bonferroni-adjustment.
Node 7 for the ‘GlaucomaM’ data is now splitted in addition (in tms).
party 0.8-3 (2006-03-20)
- ‘predict’ returned nonsensical results in case a tree was
fitted to a multivariate response variable. Now a list with one
element for each observation is returned.
party 0.8-2
- plotting buglets fixed: improved axis in MOB plots,
correct (transparent) background on jpeg() and png() devices.
party 0.8-1 (2006-02-24)
- ‘testtype’ can be Bonferroni, MonteCarlo, Univariate and
Teststatistic (instead of Raw)
- removed unnecessary constraint on weights in ‘cforest’
- maximize raw test statistics for random forest by default.
party 0.8-0 (2006-02-16)
- ‘cforest’, an implementation of Breiman's random forest ensemble
algorithm based on conditional inference trees (‘ctree’) has been
added. It is somewhat experimental and thus for advanced users only.
- ‘mob’, an object-oriented implementation of a recently suggested
algorithm for model-based recursive partitioning (Zeileis, Hothorn,
Hornik, 2005) has been added. It works out of the box for partitioning
(generalized) linear regression models, others can be easily plugged
in. A new vignette "MOB" describes the implementation ideas and provides
a few examples.
- new S3 generic reweight() with methods for (g)linearModel
for re-fitting a model with different weights.
- new argument ‘maxdepth’ to ‘ctree_control’ which allows for
growing trees of a fixed maximum depth.
- LazyData: yes
party 0.3-6 (2006-02-06)
- use a different way to deal with NAMESPACEs during in tests/*.R
following a suggestion by Prof. Brian D. Ripley
- use LazyLoad: yes instead of SaveImage: yes
party 0.3-5 (2006-02-02)
- make R CMD check happy and move src/README to inst/README
party 0.3-4
party 0.3-3 (2006-01-18)
- some new convenience functions, not exported
party 0.3-2 (2006-01-12)
- isequal must not check equality of attributes
party 0.3-1 (2005-12-02)
- fix bug introduced in version 0.3-0: coin:::factor_trafo applies
treatment contrasts in case of a binary factor but we need the full
design matrix
- documentation updates
party 0.3-0 (2005-11-29)
- ‘ytrafo’ and ‘xtrafo’ functions make use of ‘coin:::trafo’ now,
i.e. the handling of influence functions and transformations
has been unified in packages ‘party’ and ‘coin’
party 0.2-10 (2005-11-08)
- new slot ‘data’ in BinaryTree objects, slots inputnames and levels
have been removed
party 0.2-9 (2005-10-19)
- splits in categorical variables could have failed due
to weired internal checks
party 0.2-8 (2005-09-06)
- fix problems reported by codetools
party 0.2-7 (2005-08-31)
- extract inputs using menv@get("input", data = newdata) in predict
party 0.2-6 (2005-08-29)
party 0.2-5 (2005-08-19)
- ctree() now searches the global environment for variables given
in ‘formula’ when the ‘data’ argument is unspecified.
party 0.2-4 (2005-07-07)
- names of ‘criterion’ element in ‘SplittingNode’ objects were wrong
- categorical splits with ‘ctree_control(testtype = "MonteCarlo")’
were potentially wrong
party 0.2-3 (2005-06-29)
- print and plot only levels which are available in a node (for nominal
splits)
party 0.2-2 (2005-06-27)
- ./src/Makedeps will be automatically generated (on Windows)
and must not be part of the source-package.
- vignette cosmetics