Statistics Express Kurtosis inconistency Labview 2011 (Bug 215683)

[catofthefens]

Hi All,

I am currently working with a data collection program from which I need to output the skew and kurtosis values for an array of 30 data points. As I was already using the Statistics Express VI to calculate mean, s.d. etc. I just added the skew and kurtosis outputs to that VI. I then set up a basic VI to generate a 1D array of random numbers to let me compare the Express VI to the skew and kurtosis VI on the mathematics > probability and stats. palette.

The skew and kurtosis values are inconsistent between the two VIs. If I calculate the skew and kurtosis by hand I can see that the non-express VI appears to be giving the correct answer. I did some digging and found that this has been labelled as bug no. 215683 and is described as having been fixed as part of Labview 2011. I am running Labview 2011 with SP1 so I'm not quite sure why the two VIs are still inconsistent. 

I realise that I can simply use the VI which is giving the correct answer but I am curious as to why a bug which should be fixed still appears to be active in my version of Labview.

Any thoughts/ input would be greatly appreciated.

Many thanks in advance

   

[nigelchan]

Hi. Have you tried running this on a newer version of LabVIEW. This may have been repaired in a later version of LabVIEW.

[catofthefens]

Thanks for the reply, unfortunately I do not have access to a newer version of Labview. In this white paper, the bug is described as having been fixed between Labview 2010 and Labview 2011 which is why I'm a little confused as to why it is still occurring in my Labview 2011 installation.

[GerdW]

Hi cat,

why don't you get rid of that buggy (as it seems) ExpressVI?

The remaining data (StDev, variance) are provided by just another function from math->statistics palette…

You can also opne the ExpressVI to see, how it calculates those data and why it does produce different results!

[nigelchan]

Hi. I have managed to replicate the issue in LabVIEW 2013 so it appears this bug is still present. I will get in touch with our r&d team with regards to this bug and make sure this gets looked into by our developers. Are you happy to use the regular VIs in the meantime?

[catofthefens]

Hi there,

I have added the separate VIs into my program. Fortunately, as I'm fairly new to Labview, I've been checking that calculation VIs give me the correct answers so that I can understand the possibility of things like rounding errors occuring (and because it shows me how to build calculations of my own).

Thanks again for the help.

[DSPmchen]

Hi catofthefens,

The fix for bug 215683 was to add the non-express VI on statistics palette.   The express VI assumes the input sequence is a complete population to calculate the skewness and kurtosis. 

The non-express VI gives the option between population and samples.  And the default option is samples, which is a more valid option in other software.

We will add more description in the document to help user understand the difference between these VIs.

Thanks,

Michael