[SciPy-user] time series analysis

Matt Knox mattknox_ca@hotmail....
Mon Nov 5 21:47:41 CST 2007


> > Wow, this looks great. But a little complex 
> 
> > Well, one could write functions for common tasks that fascilitate it a
> > bit...

If you have any ideas for simplifying/improving things, we are certainly open 
to suggestions and would love the feedback. Being a sandbox package currently, 
there is no better time then now to get your ideas incorporated into the 
timeseries module.

> > Any idea when there will be a first binary release of all this together
> > with maskedarray?
> 
> Nope. The inclusion of maskedarray in numpy is still in the air, timeseries 
> may become a scikit sooner or later.

The timeseries package won't likely be moving anywhere until maskedarray moves 
somewhere else since it is dependent on it. I won't personally be providing 
any binaries while it is still in the sandbox either, but I am happy to 
provide advice on how to compile it on Windows.

Assuming maskedarray eventually moves into the core of numpy, my preference 
would be to put the timeseries module right into the main scipy trunk as I 
believe it is general purpose enough to warrant inclusion in scipy, and many 
people have expressed interest in it. But much discussion and debate will have 
to take place before that will happen.

Looking ahead really long term, I think built in support in matplotlib for 
TimeSeries objects (along the lines of the "plotlib" timeseries sub module) 
would be an improvement over the current approach to time series plotting in 
matplotlib too. But I have not discussed such things with any of the 
matplotlib developers and have no idea how they feel about that, and it is too 
early to discuss that yet anyway.

- Matt




More information about the SciPy-user mailing list