Compute the Spearman's rank correlation

GoogleCodeExporter commented 8 years ago

What steps will reproduce the problem?
1. Compute the Spearman's rank correlation with the same list
2. Error divide by zero

What is the expected output? What do you see instead?
Change in line 921, in lspearmanr:
    t = rs * math.sqrt((n-2) / ((rs+1.0)*(1.0-rs)))
with:
    t = rs * math.sqrt((n-2) / ((rs+1.0+TINY)*(1.0-rs+TINY)))

Adding TINY you avoid once the correlation is perfect

What version of the product are you using? On what operating system?
- 1.1

Original issue reported on code.google.com by jas...@gmail.com on 31 Aug 2008 at 8:53

GoogleCodeExporter commented 8 years ago

thanks for reporting, fixed in trunk

Original comment by istvan.a...@gmail.com on 31 Aug 2008 at 11:06

Changed state: Fixed

GoogleCodeExporter commented 8 years ago

thanks to you, if I find something else I will report to you.

Original comment by jas...@gmail.com on 31 Aug 2008 at 11:38

avatarzhang / python-statlib

Compute the Spearman's rank correlation #3