[SciPy-user] Questions on scipy.io.read_array()

charlie charlie.xia.fdu@gmail....
Thu Feb 21 15:06:37 CST 2008


Hi, David

I guess this function will work! And thanks to your suggestions.
I will take a look at these two modules and surely I will write some
feedback if they are useful to my project.

Charlie


On Thu, Feb 21, 2008 at 9:49 AM, David Huard <david.huard@gmail.com> wrote:

> Charlie,
>
> Numpy has a module called ma providing an array object that deals with
> missing values. It still lack however an official loadtxt function, but I
> worked on one a while ago. If you end up using it, I'd be grateful if you
> could provide some feedback. As for the bug in scipy.io, I think this
> function is being replaced by numpy.loadtxt.
>
> If you are dealing with time series, look at the timeseries module in
> scikits (only in SVN for now).
>
> Cheers,
>
> David
>
>
>
> 2008/2/20, charlie <charlie.xia.fdu@gmail.com>:
> >
> > Hi,
> >
> > I am a newbie to scipy.
> > I am currently using it to deal with some statistical problems with
> > possible missing values.
> > these values are labeled 'na' in my data file.
> > However when I tried to read in my data into an array and substitute
> > 'na' with -1 (for example) by:
> > read_array( datafile, ..., missing=-1)
> > The array I got doesn't cast 'na' value into -1, but 0 - the default
> > value of parameter "missing".
> > And when I check mail list, I found the issue has already be raised by
> > Joris De Ridder:
> >
> > http://article.gmane.org/gmane.comp.python.scientific.user/3700/match=read%5farray+missing
> > So I guess there is something wrong with regard to scipy.io library.
> > Does anybody come across the same problem?
> > Should I raise a ticket for this seemingly bug?
> >
> > Also, I'd like to ask for two general questions:
> > first, how efficient is python+numpy+scipy 's with major calls to
> > statistics distribution functions,
> > as compared to Matlab, C++ with CEPHES or GSL, and etc.
> > I compared it with my old R program, it seems python+numpy+scipy is
> > little bit faster.
> > Can anybody provide with some references to this?
> >
> > Another question is there a good package handle missing values well
> > within scipy?
> > Such as it can store the value as missing and fill it with different
> > inference method when desired.
> >
> > Thanks!
> >
> > Charlie
> >
> > _______________________________________________
> > SciPy-user mailing list
> > SciPy-user@scipy.org
> > http://projects.scipy.org/mailman/listinfo/scipy-user
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://projects.scipy.org/pipermail/scipy-user/attachments/20080221/e4c65aef/attachment.html 


More information about the SciPy-user mailing list