NMS PcOrd comparison with R output

[Cheryl Morgan - NOAA Affiliate]

Hi Bruce,

It's your 'old' copepod friend, Cheryl, from over at HMSC.

We are still using PcOrd to pinpoint the date of the Biological spring transition and it is part of our 'stoplight' for salmon, etc.

Some day I might retire and the younger folks will continue the timeseries, but in R. So, we finally decided to compare the output of the NMS Scores from R to that of PcOrd. The trends are very similar, so that is great. But, the range of values from PcOrd are much greater than the output from R. We have checked to see that all the settings are the same, but are unable to figure out why. 

Have you ever done this type of comparison and would you have an idea why this would be. We like the greater range of values from PCord. It has been coincidental that the Biologicals spring and fall copepod transitions happen when there is a sign change in the Axis 1 scores! This is not quite the case in the R output. Happy to send plots or data if that would help.

Thanks, Cheryl

[Bruce McCune]

Hi Cheryl, 

In our trials we obtained very similar results between NMDS in vegan and NMS in PC-ORD. But making the comparison is a little tricky because default options aren't the same. Axis scaling might have something to do with it, as might choice of the tie handling option. Also need to be certain that you're comparing solutions with the same dimensionality.

There are multiple versions of NMS in R and as I recall, some of them differ markedly from the others.

You might find interesting the attached comparison of NMS between PC-ORD and R. It is out of date and versions have changed ,so things like computation speed will be quite different now, but I suspect many of the basics haven't changed.

There are some more comparisons here...

https://www.wildblueberrymedia.net/pcord-vs-r

Bruce

--
You received this message because you are subscribed to the Google Groups "PC-ORD" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pc-ord+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pc-ord/6849cc00-8421-48ba-a238-8ecf07918d46n%40googlegroups.com.

[Cheryl Morgan - NOAA Affiliate]

Hi Bruce,

Thanks for the speedy reply. 

I didn't see any attachment, but I think you meant to add one? I'd love to see that PCORD vs R comparison.

Thanks, Cheryl

To view this discussion visit https://groups.google.com/d/msgid/pc-ord/CAK1WPPkn6BhHRyONe_-%2BFKmK%3Dqm9uDQDTyhWWhaXEtR%2BPe4LNA%40mail.gmail.com.

--

Cheryl A. Morgan

Senior Faculty Research Asst. II

Cooperative Institute for Marine Ecosystem and Resources Studies

OSU Hatfield Marine Science Center

Newport, OR 97365

[Bruce McCune]

Shoot, trying again with the attachment.

To view this discussion visit https://groups.google.com/d/msgid/pc-ord/CABwZvK%3DkXWZtKRhw3H4WZVw148DkbR4SdHOxo1uxRRZ3FOYPHw%40mail.gmail.com.

[Cheryl Morgan - NOAA Affiliate]

Hi Bruce,

Here is a plot showing what I'm talking about. This is the axis1 scores from NMS out of PCORD vs R (vegan) over our whole time series for the copepod community. The patterns match pretty well, but the peaks and troughs for PCORD are much greater than for R. We use the crossing over zero as the transition from summer to winter (or reverse) copepod communities. Any thoughts on the big difference in the range of values between the 2 softwares? The differences are not as pronounced for axis2, but since most of the variance is on axis1 that makes some sense to me.

Thoughts?

Thanks, Cheryl

To view this discussion visit https://groups.google.com/d/msgid/pc-ord/CAK1WPPkTqkeCGSuN7-OwNPjhumWYc11KdyoY3GbimJj2Xb7KoQ%40mail.gmail.com.

[Bruce McCune]

Cheryl, if you plot one set vs the other in an scatter plot is it linear? If so it is just an arbitrary scaling of the axis that makes no difference. If it is some other shape (e.g. sigmoid) then something else is going on.

Btw, it's fun to see those cyclical results -- very different from what most people are dealing with. Amazing to have those fine grained long term data.

Bruce

To view this discussion visit https://groups.google.com/d/msgid/pc-ord/CABwZvKmdH4UnpBtNHAr8%2BCHRkYDbZteR_qne-JxpC5q%2Bf4NwLA%40mail.gmail.com.