[SciPy-dev] ERROR: Tests pdist(X, 'cosine') on the Iris data set.

[Nils Wagner]

On Wed, 04 Jun 2008 01:12:55 -0700
  Damian Eads <eads at soe.ucsc.edu> wrote:
> Nils Wagner wrote:
>> Hi All,
>> 
>> Can someone reproduce the following error with
>>>>> scipy.__version__
>> '0.7.0.dev4411'
>>>>> numpy.__version__
>> '1.2.0.dev5252'
>> 
>> ======================================================================
>> ERROR: Tests pdist(X, 'cosine') on the Iris data set.
>> ----------------------------------------------------------------------
>> Traceback (most recent call last):
>>    File 
>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/cluster/tests/test_hierarchy.py", 
>> line 264, in test_pdist_cosine_iris
>>      Y_test1 = pdist(X, 'cosine')
>>    File 
>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/cluster/hierarchy.py", 
>> line 1478, in pdist
>>      dm = squareform(dm)
>>    File 
>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/cluster/hierarchy.py", 
>> line 766, in squareform
>>      raise ValueError('The distance matrix array must be 
>> symmetrical.')
>> ValueError: The distance matrix array must be 
>>symmetrical.
>> 
>> Nils
> 
> Sorry about that. This is a fairly new test in a new 
>module so I'm still 
> ironing out numerical differences between architectures. 
>Although, I was 
> not able to reproduce the error, I believe reverting 
>back to the C 
> implementation of pdist(*, 'cosine') instead of the 
>vectorized version 
> might fix the problem. I just committed the change 
>(4412).
> 
> Damian
> _______________________________________________
> Scipy-dev mailing list
> Scipy-dev at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-dev

Damian,

Thank you. Works fine for me now.

Cheers,
           Nils