Main Content

Speed Up Statistical Computations

Parallel or distributed computation of statistical functions

Statistics and Machine Learning Toolbox™ allows you to use parallel computing to speed up certain statistical computations. In parallel computing, a single MATLAB® client session distributes code segments to multiple workers for independent processing, and then combines these individual results to complete the computation. Use parallel computing to speed up resampling techniques such as bootstrap and jackknife, boosting and bagging of decision trees, cross-validation, clustering algorithms, and more. For a complete list of Statistics and Machine Learning Toolbox functions that support parallel computing, see Function List (Automatic Parallel Support).

Some functions accept gpuArray (Parallel Computing Toolbox) input arguments so that you can accelerate code by running on a graphics processing unit (GPU). For the full list of Statistics and Machine Learning Toolbox functions that accept GPU arrays, see Function List (GPU Arrays).

You must have a Parallel Computing Toolbox™ license to use the parallel computing functionality and GPU arrays.

Topics

Quick Start Parallel Computing for Statistics and Machine Learning Toolbox

Get started with parallel statistical computing.

Concepts of Parallel Computing in Statistics and Machine Learning Toolbox

Overview of the ideas in parallel statistical computations.

When to Run Statistical Functions in Parallel

Deciding when to call functions in parallel

Working with parfor

Parallel computing using parfor with statistics functions.

Implement Jackknife Using Parallel Computing

Speed up the jackknife using parallel computing.

Implement Cross-Validation Using Parallel Computing

Speed up cross-validation using parallel computing.

Implement Bootstrap Using Parallel Computing

Speed up the bootstrap using parallel computing.

Reproducibility in Parallel Statistical Computations

How to obtain identical results from repeated parallel computations.