[SciPy-user] Limits of linrgress - underflow encountered in stdtr

Robert Kern robert.kern@gmail....
Tue Jun 9 13:33:11 CDT 2009


On Tue, Jun 9, 2009 at 13:21, wierob<wierob83@googlemail.com> wrote:
> Hi,
>
>> but the p-value is
>>
>>> 6.009953e-314 and r-squared is different. While 6.009953e-314 is small
>>> enough to say its 0 and the result is highly significant
>>>
>>
>> *highly significant* ???
>>
> As far as I understood a value below a certain confidence value (usually
> 0.05 or 0.01) means the result is statistically proofed .
>> What significance level do you want to use to accept the Null when you
>> are using the result of R?
>>
>> Note: R initially only reported Pr(>|t|) < 2e-16 ***
>>
>> There are arguments for reporting any statistics only to a few
>> decimals. I wonder why?
>>
> I tested again with Scipy 0.7.0 and the p-value was 6.00995334253e-314
> the same as R's.
>
> So, for now I assume that linregress works as it should. Also, I don't
> understand why ignoring these errors is ok.

They aren't really errors in this case. That's why we give you the
ability to control how underflows are handled instead of always
raising an exception.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
  -- Umberto Eco


More information about the SciPy-user mailing list