Author |
Topic |
|
gotophysics
Germany
2 Posts |
Posted - 10/16/2015 : 05:50:13 AM
|
Hello @all,
I'm a student and work with Origin 9 Pro. At the Moment I try to distinguish different NIR-spectra with a PCA. Therefore I have 4 samples (objects) with their belonging frequency-dependind absorptionvalues (234 variables).
My Problem is that I can't perform a PCA. I put the objects (sample names) in the raws of column A(X) and the absorption values for each wavenumber in the colums B(Y) up to C234(Y):
A(X) B(Y) C1(Y) C2(Y) C3(Y) C4(Y) ... sample1 --> 236 Absorption values (spectra of sample 1) sample2 --> 236 Absorption values sample3 --> 236 Absorption values sample4 --> 236 Absorption values
So each row contains one spectra as each column contains the absorption values of one frequency of 4 different samples.
Now I'd like to perform the PCA, but I get the error message that the numbers von objects is lower than the number of variabels, so a PCA can not be performed.
But if I Change raws and colums I have a different problem. The PCA works, but the results are unphysical. The PCA assumes 4 Absorption values, belonging to one wavenumber BUT to 4 different samplespectra. I'd like to reduce the whole spectra of one sample to the principal components.
Can you give me some advises how to proceed? Thank you very much.
|
|
easwar
USA
1964 Posts |
Posted - 10/16/2015 : 3:03:33 PM
|
Hi gotophysics,
The PCA tool was improved after 9.0, and this restriction has been removed.
You can download the version 2015 evaluation version, install as OriginPro, and try with your data. Go here to download: http://originlab.com/demodownload.aspx
Easwar OriginLab |
|
|
gotophysics
Germany
2 Posts |
Posted - 10/19/2015 : 02:44:30 AM
|
Hi easwar, thank's for this advise.
|
|
|
|
Topic |
|
|
|