Orange Forum • View topic - Initial data classification

Initial data classification

A place to ask questions about methods in Orange and how they are used and other general support.

Initial data classification

Postby smmudge » Mon Jan 14, 2013 12:20

Dear Orange Community,

I am a complete newbie to Orange but have a well defined data investigation I would like to perform (MDS). All help and advice welcome.

So, I have a data table saved from Excel. The first column is the sample label and the second and subsequent columns contain my variables (continuous data). I would also like to colour the samples according to their type (location in this case).

Row one contains the header (chemical name) information,
Row two contains what? What letter / word do I use to indicate the first column is text (label)?
The colour attribute (to make it easy to visualise in the final graph) should be what? discrete?
The actual chemical data needs to be continuous?
The third row should contain nothing but the word class under the colour column? Does this sound correct?

So far, if I include colour as a header (row one), it appears as if I end up with colour being used as part of the MDS (but I do get coloured spots). If I remove the colour column, I just get grey spots but a different grouping... where am I going wrong?

Thanks. Steve

Re: Initial data classification

Postby Ales » Tue Jan 15, 2013 10:36

smmudge wrote:Row two contains what? What letter / word do I use to indicate the first column is text (label)?
Use 's' or 'string'

smmudge wrote:The colour attribute (to make it easy to visualise in the final graph) should be what? discrete?
Yes discrete.

smmudge wrote:The third row should contain nothing but the word class under the colour column? Does this sound correct?
You might also mark the label (first) column as meta (with 'm').

Re: Initial data classification

Postby smmudge » Tue Jan 15, 2013 15:52

Fantastic - I will try this and report back. Thanks.

Re: Initial data classification

Postby smmudge » Wed Jan 16, 2013 11:23

Well this all seems to work well now - it makes sense anyway. Thanks for your help here.


Return to Questions & Support



cron