help-octave
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

train_sc and test_sc


From: Andrew Cordes
Subject: train_sc and test_sc
Date: Tue, 24 May 2016 14:46:12 -0300

Hi all,

I am interested more information about these two functions: (train_sc and test_sc).
My application is Linear Discriminant Analysis.

The documentation is a bit confusing to me.

I get that the train_sc trains the classifier and that test_sc applies it.

However the documentation for train_sc seems to be jumbled together with information about test_sc and it is not clear what information is for train_sc only.  

Also while having the syntax listed is important, is there something more detailed?  Maybe some examples exist?  

Some things I didn't get from the documentation:

1. train_sc has a 'mode' and test_sc has a 'type'; and the options for these are listed, and they seem to be linked, however it is not clear (at least to me) which 'mode's and which 'type's go or don't go together, or even if some 'type's like 'MDA' combine with train_sc at all.

2. The 'mode' for test_sc has various LDA options, LD2, LDA, LD3, etc but the differences are not documented, rather hints are given.  For instance, in the train_sc documentation, LD2 points  to LDBC2 and a textbook by Duda, but a search in Duda finds nothing related to LDBC2. 

3. The train_sc returns a cluster of matrices (CC) which is not documented at all other than to say it contains "model parameters of a classifier which can be applied to test data using test_sc".  Listing the CC variable after an 'LD2' training shows that matrices CC.MD, CC.NN and CC.weights exist but nothing else.  

Thank you!

Andy

reply via email to

[Prev in Thread] Current Thread [Next in Thread]