Using a user-defined equation to create a distance matrix

hjh

Hello,

  I am trying to create a distance matrix of similarity/dissimilarity values for community data.  I would like to use a specific equation (Bray-Curtis) to construct the matrix of values; values will show similarity in multiple species abundances between multiple plots.  I am aware of the method to create a distance matrix in JMP using hierarchial clustering, but do not know how to perform a similar command using my own equation.  Is there a tool to do that, or how would I go about writing the script to do so?  

Kevin_Anderson

Hi, hjh!

 

Whatever you desire can be achieved with a script.  JSL will set you free!

 

As for how to go about writing the script to do so, I recommend a computer keyboard.

 

Seriously, though, you will need to consider several things before you plow into writing a script: how your data sources are structured and organized, the data analysis steps necessary, the output you desire, and how the script will be used.  You will probably want to write a function to implement the dissimilarity measure you choose.

 

Depending on your answers to these questions (there's a helpful checklist in Chapter 11 of the Second Edition of JSL Companion), the script could be more or less difficult to write.

 

And, if you do get to choose, I would recommend a different dissimilarity measure than Bray-Curtis.  Please reference the enclosed paper if you wish.

 

Good luck!

Chao et al Abundance Based Similarity Indices with Unseen Species Biometrics2006.pdf