Statistics Toolbox

New Features

R2014b (Version 9.1)

Released: 2 Oct 2014

Version 9.1, part of Release 2014b, includes the following enhancements:

  • Multiclass learning for support vector machines and other classifiers using the fitcecoc function
  • Generalized linear mixed-effects models using the fitglme function
  • Clustering that is robust to outliers using the kmedoids function
  • Speedup of the kmeans and gmdistribution clustering using the kmeans++ algorithm
  • Fisher's exact test for 2-by-2 contingency tables

See the Release Notes for details.

Update Now

Previous Releases

R2014a (Version 9.0) - 6 Mar 2014

Version 9.0, part of Release 2014a, includes the following enhancements:

  • Repeated measures modeling for data with multiple measurements per subject
  • fitcsvm function for enhanced performance of support vector machines (SVMs) for binary classification
  • evalclusters methods to expand the number of clusters and number of gap criterion simulations
  • p-value output from the multcompare function
  • mnrfit, lassoglm, and fitglm functions accept categorical variables as responses
  • Functions accept table inputs as an alternative to dataset array inputs
  • Functions and model properties return a table rather than a dataset array

See the Release Notes for details.

R2013b (Version 8.3) - 5 Sep 2013

Version 8.3, part of Release 2013b, includes the following enhancements:

  • Linear mixed-effects models
  • Code generation for probability distribution and descriptive statistics functions (using MATLAB Coder)
  • evaluatecluster function for estimating the optimal number of clusters in data
  • mvregress function that now accepts a design matrix even if Y has multiple columns
  • Upper tail probability calculations for cumulative distribution functions

See the Release Notes for details.

R2013a (Version 8.2) - 7 Mar 2013

Version 8.2, part of Release 2013a, includes the following enhancements:

  • Support vector machines (SVMs) for binary classification (formerly in Bioinformatics Toolbox)
  • Probabilistic PCA and alternating least-squares algorithms for principal component analysis with missing data
  • Anderson-Darling goodness-of-fit test
  • Decision-tree performance improvements and categorical predictors with many levels
  • Grouping and kernel density options in scatterhist function

See the Release Notes for details.

R2012b (Version 8.1) - 11 Sep 2012

Version 8.1, part of Release 2012b, includes the following enhancements:

  • Boosting algorithms for imbalanced data, sparse ensembles, and multiclass boosting, with self termination
  • Burr distribution for expressing a wide range of distribution shapes while preserving a single functional form for the density
  • Data import to a dataset array with the MATLAB Import Tool
  • Principal component analysis enhancements for handling NaN as missing data, weighted PCA, and choosing between EIG or SVD as the underlying algorithm
  • Speedup of k-means clustering using Parallel Computing Toolbox

See the Release Notes for details.