Need your input on numpy, Numeric, and masked array usage

From: Mary Haley <haley_at_nyahnyahspammersnyahnyah>
Date: Fri, 19 Jan 2007 10:01:56 -0700 (MST)

PyNGL Talkers,

The PyNGL team is planning to add a library of potentially hundreds
of data analysis functions to the package and we need your input.

Many of these functions are specific to climate analysis, and hence
need to be able to deal with missing values.

We've been thinking about two ways to handle these special missing

   1. Using Numeric/Numpy arrays with a separate "fill_value" variable
      to hold the missing value.

   2. Using Numeric/numpy masked arrays.

We are aware that the masked array module is undergoing some changes
in Numpy, so we're eyeing this warily to see if this is going to
change how we handle things. On top of that, we are wondering how
to handle things if people pass in a mix of Numeric and Numpy arrays.
(Do people even do this?)

We would love to hear from folks about the following:

    1. Do you deal with missing values in your Python work, and what
       do you use to handle them?

    2. Have you made the transition to Numpy? Do you use Numpy masked

    3. If you haven't made the transition to Numpy, are you
       using Numeric and/or Numeric masked arrays? Do you plan
       to move to Numpy?

    4. If you use Numeric or Numpy masked arrays, do you find them
       slow or cumbersome to use?

    5. Do you envision having to use both Numeric and Numpy arrays
       at the same time?

Thanks for your input.


pyngl-talk mailing list
Received on Fri Jan 19 2007 - 10:01:56 MST

This archive was generated by hypermail 2.2.0 : Mon Feb 05 2007 - 16:33:49 MST