Kernel density estimation in seaborn for cyclic end points

Question

I've got some data that has cyclic end points (the x-axis is longitude and hence 0 and 360 are the same point), but as far as I can tell seaborn.kdeplot doesn't have an option to specify cyclic end points.

What would be the easiest away to do univariate kernel density estimation in python/seaborn whilst accounting for the cyclic end points?

score 0 · Accepted Answer · answered Oct 22 '15 at 06:42

0

You can calculate the KDE using other libraries, and then just plot them in seaborn.

You can use the scipy KDE. It has an option to define cyclic boundary (link)

If you want more sophisticated KDE kernels (SciPy has only gaussian), you can use the scikit-learn KDE with a sphere distance metric (link).

answered Oct 22 '15 at 06:42

Elad Joseph

2,998
26
41

Sphere distance metric (Haversine Metric) requires 2D input data. What about 1D? – happy_marmoset Apr 15 '21 at 04:44
1

`PyQt-Fit` is not `scipy`. – happy_marmoset Apr 15 '21 at 05:40

Kernel density estimation in seaborn for cyclic end points

1 Answers1