cornerplot

cornerplot takes a nSamples-by-nDimensions array, and makes density plots of every combination of the dimensions, and histograms for each dimension. This is especially useful when using MCMC; you can see how the parameters in your model interact, and whether there are any tradeoffs between them.

Inspired by corner.py by Dan Foreman-Mackey.

Usage

<img src="http://wtadler.com/picdrop/cornerplot.png" width=30% align="left" />

You can obtain a plot like that pictured to the left with cornerplot(randn(500, 3)). <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br />

<img src="http://wtadler.com/picdrop/cornerplot_labels.png" width=30% align="left" />

You can also label parameters and mark true parameter values, as in cornerplot(randn(500, 3), {'a', 'b', 'c'}, randn(1, 3)). <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br />

Installation

Just put cornerplot.m in your MATLAB path. You also need to install the free kernel density estimator kde2d by Zdravko Botev.