Statistics Toolbox 7.6

Latest Features

Version 7.6

Released: 1 Sep 2011
 

Version 7.6, part of Release 2011b, includes the following enhancements:

• Lasso and elastic net regularization for shrinkage estimation and variable selection with linear regression
• Linear and quadratic discriminant analysis object, compatible with the classification framework
• Nearest neighbor searching for points within a fixed distance
• datasample function for weighted or unweighted sampling of data, with or without replacement

See the Release Notes for details.

Update Now

• Download the latest version if your software maintenance service is current.
• Purchase or request a renewal quote if your maintenance service has lapsed.
• Try the latest version by downloading or requesting trial software.

---

Version 7.5

Released: 8 Apr 2011
 

Version 7.5, part of Release 2011a, includes the following enhancements:

• Boosted decision trees for classification and regression
• Memory and performance improvements in hierarchical clustering
• Tie detection in kth nearest neighbor searching (knnsearch function)
• Weighted residuals in nlmefit function

See the Release Notes for details.

Version 7.4

Released: 3 Sep 2010
 

Version 7.4, part of Release 2010b, includes the following enhancements:

• Parallel computing support for the following functions: plsregress, sequentialfs, nnmf, rowexch, candexch, and cordexch when used with Parallel Computing Toolbox
• New relieff feature ranking algorithm for correlated features
• Enhanced nlmefit algorithm that supports constant, proportional, combined, and exponential error models
• Surrogate splits option for decision trees that handles missing data, estimates variable importance, and unmasks variables
• Enhanced dfittool that saves fits to the workspace as distribution objects

See the Release Notes for details.

Version 7.3

Released: 5 Mar 2010
 

Version 7.3, part of Release 2010a, includes the following enhancements:

• New stochastic algorithm for fitting NLME models, with parameter transformations, support for nonconstant error variance, and improved convergence
• New functions for k-nearest neighbor (kNN) search to efficiently find the closest points to any query point
• New option in perfcurve function to compute confidence intervals for classifier performance curves
• New options to weight resampling probabilities to broaden the range of models supported by bootstrp, bootci, and jackknife functions

See the Release Notes for details.

Version 7.2

Released: 4 Sep 2009
 

Version 7.2, part of Release 2009b, includes the following enhancements:

• New parallel execution support for the bootstrp, bootci, crossval, and jackknife functions, and the TreeBagger class
• Conversion between tall and wide dataset arrays
• New capability in nlmefit function for monitoring or cancelling calculations
• New ability to read and write SAS^® Transport (.xpt) files

See the Release Notes for details.