Forum Home  >  perClass  >  Bug forum  >  Thread
Search  
  
  Advanced Search
Welcome Guest   ·   Login   ·   Register   ·   Member List
Username Remember Me?
Password   forgot password?
   
   
Trying to use sdscatterd with a Prtools dataset
 
Mariano Llamedo Soria
Posted: 03 June 2010 04:50 PM   [ Ignore ]  
Jr. Member
RankRank
Total Posts:  41
Joined  2008-07-11

I have many datasets generated with Prtools and the first thing I tried in PRSD was visualizing the data with sdscatter. I found the following error:

>> sdscatter(dsTarget_dataset)

PRSD Studio Lite 2.2.1 (03-May-2010), Copyright (C) 2007-2010, PR Sys Design, All rights reserved
Only for non-commercial use.

??? Index exceeds matrix dimensions.

Error in ==> D:\Mariano\papers\ECG Classification\scripts\common\prsd_toolbox\@sddata\private\prtools_import.p>prtools_import at 12


Error in ==> D:\Mariano\papers\ECG Classification\scripts\common\prsd_toolbox\@sddata\sddata.p>sddata at 56


Error in ==> D:\Mariano\papers\ECG Classification\scripts\common\prsd_toolbox\sdscatter.p>sdscatter at 38

 
>> 
>> 
>> struct(dsTarget_dataset)

ans = 

        data: [100695x23 double]
     lablist: {6x4 cell}
        nlab: [100695x5 double]
     labtype: 'crisp'
     targets: []
     featlab: [23x1 double]
     featdom: {[]  []  []  []  []  []  []  []  []  []  []  []  []  []  []  []  []  []  []  []  []  []  [2x2 double]}
       prior: []
        cost: []
     objsize: 100695
    featsize: 23
       ident: [1x1 struct]
     version: {[1x1 struct]  '02-Jun-2010 11:30:09'}
        name: 'MITBIH-AR-2 leads'
        user: [1x1 struct]

Am I doing anything wrong ?

Thanks,
Mariano.

Profile
 
 
pavel
Posted: 03 June 2010 05:00 PM   [ Ignore ]   [ # 1 ]  
Administrator
Avatar
RankRankRankRank
Total Posts:  240
Joined  2008-04-26

Hi Mariano,

the problem seems to be lablist-related.  Can you post the result of:

>> s=struct(dsTarget_dataset)
>> s.lablist

Pavel

Profile
 
 
Mariano Llamedo Soria
Posted: 03 June 2010 05:11 PM   [ Ignore ]   [ # 2 ]  
Jr. Member
RankRank
Total Posts:  41
Joined  2008-07-11

>> s=struct(dsTarget_dataset);
>> s.lablist

ans = 

    [14x46 char]     []              []    'crisp'
    [ 6x16 char]     []              []    'crisp'
    [ 5x16 char]     []              []    'crisp'
    [ 4x9  char]     []              []    'crisp'
    [ 3x8  char]     []              []    'crisp'
    [ 5x7  char]    [2]    [1x5 double]         []

I attached the this variable in case you need to inspect it. It is a mat file, but I must rename it because the forum does not allow me to upload .mat files.

M.

File Attachments
matlab.txt  (File Size: 1KB - Downloads: 93)
Profile
 
 
pavel
Posted: 03 June 2010 06:09 PM   [ Ignore ]   [ # 3 ]  
Administrator
Avatar
RankRankRankRank
Total Posts:  240
Joined  2008-04-26

Hi Mariano,

the problem was related to multiple sets of labels in your data set. I’m attaching the patch. Just copy the file in prsd_toolbox\@sddata\private directory.

PRSD Studio Lite is limited to 300 samples and 3 classes so you will have to scale down your problem. If you wish to test the full (academic?) version, let me know at

Does it help?

Pavel

File Attachments
prtools_import.p  (File Size: 1KB - Downloads: 85)
Profile
 
 
   
 
 
‹‹ Problem with sddecide      Problem with untrained classifiers ››