Statistics and probability

Describe samples, fit distributions, test differences and run Monte Carlo studies with mean, fitdist, ttest2, bootstrp, fitlm and the other statistics functions you already know.

  • Existing MATLAB statistics code runs as written
  • Distribution fits, tests, bootstrap and regression, with no toolbox to buy
  • Random numbers and reductions run on the GPU for large simulations
  • Free to try, with nothing to install
Download RunMat

Explore the RunMat runtime

A Monte Carlo estimate settling inside a narrowing two-standard-error band, drawn by RunMat from the saved analysis study
A million-draw tolerance stack-up, drawn by RunMat. The estimate settles as its two-standard-error band narrows with the square root of the draws.

Why use RunMat for statistics

No toolbox to buy for fitdist and ttest2

Fitting a distribution or running a t-test usually means paying for a statistics toolbox, or waiting for a shared license. fitdist, ttest2, bootstrp, fitlm and the distribution functions are part of the RunMat runtime next to mean and histogram, free for individuals, education, research and smaller companies.

Runtime

Run far more draws in the same time

Simulation studies are usually limited by how many draws fit in the time available. rand, randn, mean, std and cumsum run on the GPU when the arrays are large, so the same vectorized script runs far more draws and gives tighter intervals.

GPU

Get the same p-value on every rerun

A result that changes on each run can't be checked by anyone else. rng(seed) repeats a run exactly in RunMat, and each run keeps its script, data and figures.

Desktop app

Let a reviewer rerun the analysis

A reviewer can open the same .m file in the browser with nothing to install and get the same numbers from the same seed, or open the project on the desktop alongside you.

Collaboration

Examples

A clearance histogram with the interference line above the running estimate inside its standard-error band, rendered by RunMat

Monte Carlo tolerance stack-up

A shaft in a bore: interference probability, its standard error and when to stop.

The error of a Monte Carlo estimate of pi against the number of points, beside a one-over-root-N line, rendered by RunMat

Estimate pi

The smallest Monte Carlo study, with its error at every sample size.

Side-by-side box plots of two batches with the t-test result in the title, rendered by RunMat

Do two batches differ?

Box plots, a Welch t-test and the interval on the difference.

A histogram with its fitted normal curve beside a QQ plot, rendered by RunMat

Fit a distribution

A normal fit over the histogram, a QQ plot and a KS test.

The bootstrap distribution of a median with the 95% interval marked, rendered by RunMat

Bootstrap a median

Two thousand resamples of a skewed sample and a percentile interval.

A regression line with its prediction band beside residuals against fitted values, rendered by RunMat

Regression with diagnostics

A fitted line, a prediction interval and the residuals that judge it.

Box plots of three suppliers with the ANOVA result in the title, rendered by RunMat

Compare three suppliers

One-way ANOVA across three groups, with the F test and group means.

A histogram of cycle times with a kernel density curve showing two peaks, rendered by RunMat

Smooth a histogram

A kernel density estimate that finds the peaks a histogram blurs.

Variance explained by each principal component beside the scores on the first two, rendered by RunMat

Principal components

Which directions carry the variance in four correlated sensors.

Try it on your own data

Download RunMat
Read the docs