[Scipy-tickets] [SciPy] #835: nan propagation in stats.distribution

SciPy scipy-tickets@scipy....
Wed Feb 11 22:01:33 CST 2009


#835: nan propagation in stats.distribution
-------------------------+--------------------------------------------------
 Reporter:  josefpktd    |        Owner:  somebody
     Type:  defect       |       Status:  new     
 Priority:  normal       |    Milestone:  0.8.0   
Component:  scipy.stats  |      Version:  devel   
 Severity:  normal       |   Resolution:          
 Keywords:               |  
-------------------------+--------------------------------------------------
Changes (by josefpktd):

  * component:  Other => scipy.stats

Comment:

 All the bounds (and distribution parameter) checking is already done in
 stats.distributions but they need to be rearranged to propagate nan.
 Currently, the output array is initialized with zeros and then filled with
 valid values, instead it should be initialized with nan (`Badvalue`) and
 then filled with valid values including the zeros where they are correct.
 So there is no additional performance cost compared to the current
 implementation.

 The use case I had, was, that in the t-tests, the zero-division error
 might produce a nan. In this case the t-statistic is nan and I would like
 to also automatically get a nan for the p-value.

-- 
Ticket URL: <http://scipy.org/scipy/scipy/ticket/835#comment:3>
SciPy <http://www.scipy.org/>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list