turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- Re: How to organize multidimensional spectral data for partial least squares (PL...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 25, 2015 6:44 AM
(1175 views)

I have a large data set that has multiple independent and dependent variables. For smaller systems, I have organized the data such that the independent variables are defined in the Column Name. However, for much larger sets, where more than a few Xs and multiple Ys are involved, this becomes an extremely tedious exercise. Moreover, it makes for very wide, unwieldy tables and is hard to visualize graphically.

Does anyone in the community have any suggestions on a more straightforward method of organizing large spectral data sets for PLS?

Thanks.

2 REPLIES

Highlighted
##
##### Re: How to organize multidimensional spectral data for partial least squares (PLS)?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Mar 8, 2015 11:19 AM
(1037 views)

I'm not sure what kind of suggestions you're looking for but perhaps creating column groups could be helpful.

Select a set of columns and choose **Col -> Group Columns**.

-Jeff

-Jeff

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Mar 9, 2015 5:27 AM
(1037 views)

Jeff,

Thanks for the response. I know about organizing data via column groups; what I'm looking for is a more organized way to input multi-spectral input to PLS. For PLS input, I currently convert all the independent variables to columns using 'split' operations and then have the dependent variable(s) as the column value. It would be great if PLS could accept x,y,z, ... as values in their respective columns, and then have the response (dependent) variables as separate columns.

As for this kind of data being hard to visualize, it would be great to (at least) have a graphics capability which would allow topographical maps to be generated from 3D data. 3D scatterplot is just okay for this kind of visualization.