[SciPy-user] fill a timeseries with masked data by correlation from another series

[Marco Tuckner]

> Mmh, that'll depend on what you want to do. More specific questions would be 
> better answered... 
That was meant as a introduction to the questions that followed ;-)

>> I would like to correlate two (or more timeseries) to estimate invalid and
>> masked values in one series based on the values of another complete series
>> using the correlation coefficient.
>> How can I to that?
>>
>> I tried this without successs:
>> ### CODE ###
>> In [22]: scipy.stats.corrcoef(series,modseries)
> 
> scipy.stats.corrcoef doesn't accept masked arrays. More exactly, it transforms 
> a masked array into a regular ndarray, therefore losing the mask. Most of the 
> functions of scipy.stats work that way. I'm currently rewriting (most of) 
> them to support masked array,
Good news!

> hopefully I'll be able to post something in the 
> next few days. Where should I post them ? In scipy.stats.mstats ? In 
> numpy.ma ?
I don't know because I am not a developer. But I would use scipy.stats.mstats.
 
>> Furthermore I like to calculate other tests like t-test, chi-test,
>> Kolmogorov-Smirnof on the two sets and get distribution parameters (RMSE,
>> MBE, STD) for each series.
>> How can this be done?
Actually, I was able to compute Kolmogorov-Smirnof (KSI). But correlation wasn't
possible.
> 
> For the basic statistical tests in scipy.stats, see above.
> For STD, you can use the std method
> For RMSE, you could try to sum the squares of the anomalies

> For MBE, you're losing me: Member of the British Empire
Some publications call it Mean Bias Error, I think it's the same as
http://en.wikipedia.org/wiki/Mean_squared_error

Thanks a lot for your efforts!