F-Test P-Value for Two-Tailed test is >1

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Variances and Degrees of Freedom a calculated from the attached data sets 
Sample 1 (Random_50_0.5.txt) has a variance of 1.05766555271071 and DoF of 49. 
and Sample 2 (Random_50_0.7.txt) has a variance of 1.16570834301777 and DoF of 
49. 

2. Enter data in Constructor FTest(Sample1.Variance, Sample2.Variance, 
Sample1.DoF, Sample2.DoF, TwoSampleHypothesis.ValuesAreDifferent);

3. Compare these with One Tailed Tests using Constructors: 
One Tailed - Greater:
FTest(Sample1.Variance, Sample2.Variance, 
Sample1.DoF, Sample2.DoF, TwoSampleHypothesis.FirstValueIsGreaterThanSecond);
One Tailed - Less:
FTest(Sample1.Variance, Sample2.Variance, 
Sample1.DoF, Sample2.DoF, TwoSampleHypothesis.FirstValueIsSmallerThanSecond));

What is the expected output? What do you see instead?

Results: 
Two Tailed p-value: 1.26772493058357
One Tail Greater p-value: 0.633862465291784
One Tail Less p-value: 0.366137534708217

For the two tailed test I would have expected 0.732275069416434 (i.e. double 
the One Tail Less, this value was confirmed by R and SciPy).

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

Windows8 (Visual Studio 2008), 64.bit, Accord.Net Version 2.8.0

Please provide any additional information below.

Original issue reported on code.google.com by andrewhd...@gmail.com on 20 Dec 2012 at 10:24

Attachments:

• Random_50_0.5.txt
• Random_50_0.7.txt

[GoogleCodeExporter]

Thanks, will be fixed on next release.

Original comment by cesarso...@gmail.com on 20 Dec 2012 at 2:45

• Changed state: Accepted

[GoogleCodeExporter]

Fixed on release 2.8.1

Original comment by cesarso...@gmail.com on 20 Dec 2012 at 10:02

• Changed state: Fixed