Return to the 'statistics' package
Summary of important user-visible changes for statistics 1.4.3:
-------------------------------------------------------------------
New functions:
==============
** anova1 (patch #10127)
kruskalwallis
** cluster (patch #10009)
** clusterdata (patch #10012)
** confusionchart (patch #9985)
** confusionmat (patch #9971)
** cophenet (patch #10040)
** datasample (patch #10050)
** evalclusters (patch #10052)
** expfit (patch #10092)
explike
** gscatter (patch #10043)
** ismissing (patch #10102)
** inconsistent (patch #10008)
** mhsample.m (patch #10016)
** ncx2pdf (patch #9711)
** optimalleaforder.m (patch #10034)
** pca (patch #10104)
** rmmissing (patch #10102)
** silhouette (patch #9743)
** slicesample (patch #10019)
** wblplot (patch #8579)
Improvements:
=============
** anovan.m: use double instead of toascii (bug #60514)
** binocdf: new option "upper" (bug #43721)
** boxplot: better Matlab compatibility; several Matlab-compatible
plot options added (OutlierTags, Sample_IDs, BoxWidth, Widths,
BoxStyle, Positions, Labels, Colors) and an Octave-specific one
(CapWidhts); demos added; texinfo improved (patch #9930)
** auto MPG (carbig) sample dataset added from
https://archive.ics.uci.edu/ml/datasets/Auto+MPG (patch #10045)
** crosstab.m: make n-dimensional (patch #10014)
** dendrogram.m: many improvements (patch #10036)
** fitgmdist.m: fix typo in ComponentProportion (bug #59386)
** gevfit: change orientation of results for Matlab compatibility (bug #47369)
** hygepdf: avoid overflow for certain inputs (bug #35827)
** kmeans: efficiency and compatibility tweaks (patch #10042)
** pdist: option for squared Euclidean distance (patch #10051)
** stepwisefit.m: give another option to select predictors (patch #8584)
** tricdf, triinv: fixes (bug #60113)
Summary of important user-visible changes for statistics 1.4.2:
-------------------------------------------------------------------
** canoncorr: allow more variables than observations
** fitgmdist: return fitgmdist parameters (Bug #57917)
** gamfit: invert parameter per docs (Bug #57849)
** geoXXX: update docs 'number of failures (X-1)' => 'number of failures (X)' (Bug #57606)
** kolmogorov_smirnov_test.m: update function handle usage from octave6+ (Bug #57351)
** linkage.m: fix octave6+ parse error (Bug #57348)
** unifrnd: changed unifrnd(a,a) to return a 0 rather than NaN (Bug #56342)
** updates for usage of depreciated octave functions
Summary of important user-visible changes for statistics 1.4.1:
-------------------------------------------------------------------
** update install scripts for octave 5.0 depreciated functions
** bug fixes to the following functions:
pdist2.m: use max in distEucSq (Bug #50377)
normpdf: use eps tolerance in tests (Bug #51963)
fitgmdist: fix an output bug in fitgmdist
t_test: Set tolerance on t_test BISTS (Bug #54557)
gpXXXXX: change order of inputs to match matlab (Bug #54009)
bartlett_test: df = k-1 (Bug #45894)
gppdf: apply scale factor (Bug #54009)
gmdistribution: updates for bug #54278, ##54279
wishrnd: Bug #55860
Summary of important user-visible changes for statistics 1.4.0:
-------------------------------------------------------------------
** The following functions are new:
canoncorr
fitgmdist
gmdistribution
sigma_pts
** The following functions have been moved from the statistics package but are
conditionally installed:
mad
** The following functions have been moved from octave to be conditionally
installed:
BASE
cloglog
logit
prctile
probit
qqplot
table (renamed to crosstab)
DISTRIBUTIONS
betacdf
betainv
betapdf
betarnd
binocdf
binoinv
binopdf
binornd
cauchy_cdf
cauchy_inv
cauchy_pdf
cauchy_rnd
chi2cdf
chi2inv
chi2pdf
chi2rnd
expcdf
expinv
exppdf
exprnd
fcdf
finv
fpdf
frnd
gamcdf
gaminv
gampdf
gamrnd
geocdf
geoinv
geopdf
geornd
hygecdf
hygeinv
hygepdf
hygernd
kolmogorov_smirnov_cdf
laplace_cdf
laplace_inv
laplace_pdf
laplace_rnd
logistic_cdf
logistic_inv
logistic_pdf
logistic_rnd
logncdf
logninv
lognpdf
lognrnd
nbincdf
nbininv
nbinpdf
nbinrnd
normcdf
norminv
normpdf
normrnd
poisscdf
poissinv
poisspdf
poissrnd
stdnormal_cdf
stdnormal_inv
stdnormal_pdf
stdnormal_rnd
tcdf
tinv
tpdf
trnd
unidcdf
unidinv
unidpdf
unidrnd
unifcdf
unifinv
unifpdf
unifrnd
wblcdf
wblinv
wblpdf
wblrnd
wienrnd
MODELS
logistic_regression
TESTS
anova
bartlett_test
chisquare_test_homogeneity
chisquare_test_independence
cor_test
f_test_regression
hotelling_test
hotelling_test_2
kolmogorov_smirnov_test
kolmogorov_smirnov_test_2
kruskal_wallis_test
manova
mcnemar_test
prop_test_2
run_test
sign_test
t_test
t_test_2
t_test_regression
u_test
var_test
welch_test
wilcoxon_test
z_test
z_test_2
** Functions marked with known test failures:
grp2idx: bug #51928
gevfir_lmom: bug #31070
** Other functions that have been changed for smaller bugfixes, increased
Matlab compatibility, or performance:
dcov: returned dcov instead of dcor. added demo.
violin: can be used with subplots. violin quality improved.
princomp: Fix expected values of tsquare in unit tests
fitgmdist: test number inputs to function
hist3: fix removal of rows with NaN values
** added the packages test data to install
Summary of important user-visible changes for statistics 1.3.0:
-------------------------------------------------------------------
** The following functions are new:
bbscdf bbsinv bbspdf bbsrnd
binotest
burrcdf burrinv burrpdf burrrnd
gpcdf gpinv gppdf gprnd
grp2idx
mahal
mvtpdf
nakacdf nakainv nakapdf nakarnd
pdf
tricdf triinv tripdf trirnd
violin
** Other functions that have been changed for smaller bugfixes, increased
Matlab compatibility, or performance:
betastat
binostat
cdf
combnk
gevfit
hist3
kmeans
linkage
randsample
squareform
ttest
Summary of important user-visible changes for statistics 1.2.4:
-------------------------------------------------------------------
** Made princomp work with nargout < 2.
** Renamed dendogram to dendrogram.
** Added isempty check to kmeans.
** Transposed output of hist3.
** Converted calculation in hmmviterbi to log space.
** Bug fixes for
stepwisefit wishrnd.
** Rewrite of cmdscale for improved compatibility.
** Fix in squareform for improved compatibility.
** New cvpartition class, with methods:
display repartition test training
** New sample data file fisheriris.txt for tests
** The following functions are new:
cdf crossval dcov pdist2 qrandn randsample signtest ttest ttest2
vartest vartest2 ztest
Summary of important user-visible changes for statistics 1.2.3:
-------------------------------------------------------------------
** Made sure that output of nanstd is real.
** Fixed second output of nanmax and nanmin.
** Corrected handle for outliers in boxplot.
** Bug fix and enhanced functionality for mvnrnd.
** The following functions are new:
wishrnd iwishrnd wishpdf iwishpdf cmdscale
Summary of important user-visible changes for statistics 1.2.2:
-------------------------------------------------------------------
** Fixed documentation of dendogram and hist3 to work with TexInfo 5.
Summary of important user-visible changes for statistics 1.2.1:
-------------------------------------------------------------------
** The following functions are new:
pcares pcacov runstest stepwisefit hist3
** dendogram now returns the leaf node numbers and order that the nodes were displayed in.
** New faster implementation of princomp.
Summary of important user-visible changes for statistics 1.2.0:
-------------------------------------------------------------------
** The following functions are new:
regress_gp dendogram plsregress
** New functions for the generalized extreme value (GEV) distribution:
gevcdf gevfit gevfit_lmom gevinv gevlike gevpdf gevrnd gevstat
** The interface of the following functions has been modified:
mvnrnd
** `kmeans' has been fixed to deal with clusters that contain only
one element.
** `normplot' has been fixed to avoid use of functions that have been
removed from Octave core. Also, the plot produced should now display some
aesthetic elements and appropriate legends.
** The help text of `mvtrnd' has been improved.
** Package is no longer autoloaded.
Summary of important user-visible changes for statistics 1.1.3:
-------------------------------------------------------------------
** The following functions are new in 1.1.3:
copularnd mvtrnd
** The functions mnpdf and mnrnd are now also usable for greater numbers
of categories for which the rows do not exactly sum to 1.
Summary of important user-visible changes for statistics 1.1.2:
-------------------------------------------------------------------
** The following functions are new in 1.1.2:
mnpdf mnrnd
** The package is now dependent on the io package (version 1.0.18 or
later) since the functions that it depended of from miscellaneous
package have been moved to io.
** The function `kmeans' now accepts the 'emptyaction' property with
the 'singleton' value. This allows for the kmeans algorithm to handle
empty cluster better. It also throws an error if the user does not
request an empty cluster handling, and there is an empty cluster.
Plus, the returned items are now a closer match to Matlab.
Summary of important user-visible changes for statistics 1.1.1:
-------------------------------------------------------------------
** The following functions are new in 1.1.1:
monotone_smooth kmeans jackknife
** Bug fixes on the functions:
normalise_distribution combnk
repanova
** The following functions were removed since equivalents are now
part of GNU octave core:
zscore
** boxplot.m now returns a structure with handles to the plot elemenets.
Summary of important user-visible changes for statistics 1.1.0:
-------------------------------------------------------------------
** IMPORTANT note about `fstat' shadowing core library function:
GNU octave's 3.2 release added a new function `fstat' to return
information of a file. Statistics' `fstat' computes F mean and
variance. Since MatLab's `fstat' is the equivalent to statistics'
`fstat' (not to core's `fstat'), and to avoid problems with the
statistics package, `fstat' has been deprecated in octave 3.4
and will be removed in Octave 3.8. In the mean time, please
ignore this warning when installing the package.
** The following functions are new in 1.1.0:
normalise_distribution repanova combnk
** The following functions were removed since equivalents are now
part of GNU octave core:
prctile
** The __tbl_delim__ function is now private.
** The function `boxplot' now accepts named arguments.
** Bug fixes on the functions:
harmmean nanmax nanmin regress
** Small improvements on help text.
Package: statistics