Summary of important user-visible changes for statistics 1.5.0:
-------------------------------------------------------------------

 Important Notice: 1) dependency change to Octave>=6.1.0
                   2) `mean` shadows core Octave's respective function
                   3) removed dependency on `io` package
                   4) incompatibility with the `nan` package
 New functions:
 ==============

 ** anova2 (fully Matlab compatible)

 ** bvncdf

 ** cdfcalc, cdfplot

 ** chi2gof (fully Matlab compatible. bug #46764)

 ** chi2test (bug #58838)

 ** cholcov (fully Matlab compatible)

 ** ecdf (fully Matlab compatible)

 ** evfit (fully Matlab compatible)

 ** evlike (fully Matlab compatible)

 ** f2fx (missing function: bug #48146)

 ** fitlm (mostly Matlab compatible)

 ** fillmissing (patch #10102)

 ** friedman (fully Matlab compatible)

 ** grpstats (complementary to manova1)

 ** kruskalwallis (fully Matlab compatible)

 ** kstest (fully Matlab compatible)

 ** kstest2 (fully Matlab compatible. bug #56572)

 ** libsvmread, libsvmwrite (I/O functions for LIBSVM data files)

 ** manova1 (fully Matlab compatible)

 ** manovacluster (fully Matlab compatible)

 ** mean (fully Matlab compatible, it shadows mean from core Octave)

 ** multcompare (fully Matlab compatible)

 ** mvtcdfqmc

 ** ranksum (fully Matlab compatible. bug #42079)

 ** standardizeMissing (patch #10102)

 ** svmpredict, svmtrain (wrappers for LIBSVM 3.25)

 ** tiedrank (complementary to ranksum)

 Improvements:
 =============

 ** anova1: added extra feature for performing Welch's ANOVA (PR #15)

 ** anovan: mostly Matlab compatible, extra features. (patch #10123, PR #1-42)

 ** binopdf: implement high accuracy Loader algorithm for m>=10 (bug #34362)

 ** cdf: extended to include all available distributions

 ** crosstab: can handle char arrays, fixed ordering of groups

 ** gaminv: fixed accuracy for small 1st argument (bug #56453)

 ** geomean: fully Matlab compatible. (patch #59410)

 ** gevlike: fully Matlab compatible. ACOV output is the inverse Fisher Inf Mat

 ** grp2idx (fully Matlab compatible, indexes in order of appeearance)

 ** harmmean: fully Matlab compatible.

 ** hygepdf: added optional parameter "vectorexpand" to facilitate
    vectorization of other hyge functions. Allows different inputs lengths
    for x and t,m,n parameters, with broadcast expanded output (bug #34363)

 ** hygecdf: improved vectorization for non-scalar inputs.
    hygeinv
    hygernd

 ** ismissing: corrects handling of n-D arrays, NaN indicators, and improves
    matlab compatibility for different data types. (patch #10102)

 ** kmeans: improved help file, evaluate efficiency (bug #8959)

 ** laplace_cdf: allow for parameters mu and scale (bug #58688)
    laplace_inv
    laplace_pdf
    logistic_cdf
    logistic_inv
    logistic_pdf

 ** logistic_regression: fixed incorrect results (bug #60348)

 ** mvncdf: improved performance and accuracy (bug #44130)

 ** normplot: fixed ploting error (bug #62394), updated features

 ** pdf: extended to include all available distributions

 ** pdist: updated the 'cosine' metric to be more efficient (bug #62495)

 ** rmmissing: corrects cellstr array handling and
    improves matlab compatibility for different data types. (patch #10102)

 ** signtest: fix erroneous results, fully Matlab compatible (bug #49961)

 ** ttest2: can handle NaN values as missing data (bug #58697)

 ** violin: fix parsing color vector affecting Octave>=6.1.0 (bug #62805)

 ** wblplot: fixed coding style and help texinfo. (patch #8579)

 Removed Functions:
 ==================

 ** anova (replaced by anova1)

 ** caseread, casewrite (do not belong here)

 ** chisquare_test_homogeneity (replaced by chi2test)

 ** chisquare_test_independence (replaced by chi2test)

 ** kolmogorov_smirnov_test (replaced by kstest)

 ** kolmogorov_smirnov_test_2 (replaced by kstest2)

 ** kruskal_wallis_test (replaced by kruskalwallis)

 ** manova (replaced by manova1)

 ** repanova (replaced by anova2)

 ** sign_test (replaced by updated signtest)

 ** tblread, tblwrite (belong to `io` package when tables are implemented)

 ** t_test, t_test_2 (deprecated: use ttest & ttest2)

 ** wilcoxon_test (replaced by ranksum)

 Available Data Sets:
 ====================

 ** acetylene         Chemical reaction data with correlated predictors
 ** arrhythmia        Cardiac arrhythmia data from the UCI machine learning repository
 ** carbig            Measurements of cars, 1970–1982
 ** carsmall          Subset of carbig. Measurements of cars, 1970, 1976, 1982
 ** cereal            Breakfast cereal ingredients
 ** examgrades        Exam grades on a scale of 0–100
 ** fisheriris        Fisher's 1936 iris data
 ** hald              Heat of cement vs. mix of ingredients
 ** heart_scale.dat   Used for SVM testing
 ** kmeansdata        Four-dimensional clustered data
 ** mileage           Mileage data for three car models from two factories
 ** morse             Recognition of Morse code distinctions by non-coders
 ** popcorn           Popcorn yield by popper type and brand
 ** stockreturns      Simulated stock returns
 ** weather           Daily high temperatures in the same month in two consecutive years
