search

srikanthguduru / imrmc

Automatically exported from code.google.com/p/imrmc
0 stars 0 forks source link

No statistical analysis results for a user #17

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
What steps will reproduce the problem?
1. Open file, select modalities, click on "MRMC variance analysis"
2. AUCs print, but none of the rest of the statistical results print
3.

---
What is the expected output?
All the statistical analyses

---
What do you see instead?

---
What is the recommended fix?

---
Please use labels and text to provide additional information.
Windows 7
java version 8 update 25

Original issue reported on code.google.com by Brandon.Gallas on 17 Nov 2014 at 10:06

GoogleCodeExporter commented 9 years ago 
We figured out the problem. We were lucky enough to replicate your problem on 
another computer and saw the error messages sent to the console. It is not a 
java version or package issue.

The data you were using yielded ms_tr=0.0 because the readers aucs were 
identical to each other and identical to the average. The ms_tr is in the 
denominator of the Hillis degrees of freedom estimate. So DF_Hillis was 
infinity. Apparently, some versions of java (or computers) handle infinity 
better than others, not sure exactly how or why.

Anyhow, when ms_tr=0.0, we are adding alerts/flags to the software and setting 
DF_Hillis to 50. The user will know there is a problem with the Hillis degrees 
of freedom and will know the steps taken to continue and produce results. 

Do you need a pre-release update that better handles the infinite DF_Hillis?

Original comment by Brandon.Gallas on 18 Nov 2014 at 4:16