Data points of value 0 collapse onto a single point.

[bmahmou]

Hi, Thank you very much for your wonderful matlab package.

This is not so much of a bug/issue but a nuisance which I thought more people should be aware of, which might be worth putting in the documentation.

It appears that data points that have value 0 are overlayed on top of each other when plotted (unless all data points have the value 0 ,in which case all individual points are shown). This can sometimes result in misinterpretation of the number of data points plotted.
For examples, I have attached some figures below. Please look at category "mouth"
In all figures, the number of data points are the same in category "mouth".

• allNonzero example all data points are equal and non zero (0.3)
• allZero example has all data points are equal and 0.
• moreThanOneNonZero has data that are all equal either at 0.3 or 0
• onNonZero has data that are all zeros except one data point at 0.3

As illustrated this can be a bit misleading about the number of points that are actually plotted. Just wanted to point this out.
Thank you for your time.

[bastibe]

Indeed the points seem to converge on zero. How strange. From a cursory look at the code, I don't see an obvious reason for that behavior. And I don't have access to Matlab any longer, so I can't actually debug it myself.

I would expect that there's a subtle bug in the jitterstrength calculation. I'd be grateful for a pull request that fixes the issue.

(The case of the all-zero values is handled separately and drawn as a line, since a kernel density estimate makes no sense if there's only one value.)