Paired T-Test with Uneven Sample Population

KNIME Analytics Platform
1

Hi Knime Forum

I am trying to use the Paired t-test node (under Statistics) to run on uneven sample sizes.  My data looks like the following (the numbers are made up)

 

Sample 1 Sample 2
34.5 ?
52.4 ?
44.3 ?
? 8.4
? 9.2
? 7.4
? 5.4
? 8.4
? 7.2
? 7.8

KNIME gives me the following "Execute failed: degrees of freedom (-1)"

I understand that using uneven sample sizes in a t-test requires a little more work than a equal sample size, is KNIME unable to handle this?  Should I look into using an R script here?  Thank you for the help.


Dennis 

 

2

Quick update, 

I tried some test data in the following format and the T-test worked ok (I import the data via file)

Sample 1 Sample 2
34.5 8.4
52.4 9.2
44.3 7.4
? 5.4
? 8.4
? 7.2
? 7.8

The first post the data was generate in KNIME via two different methods and then "concatentated".   Is there a good way to convert the data in the initial post to how it looks in this post?
 

Thank you


Dennis

3

Ok I can answer my own question...hopefully down the road someone will find this useful.

Rather than concatentate this data after processing, I process all the data upfront and then Pivot the output with each sample generating a LIST in the aggregation method.  I then use an Ungroup node to generate data as above.  

Dennis

4