[MATRIX-SIG] Much ado about nothingness.

[Jim Hugunin]

I think that Zane did a great job of summarizing the intentions of NumPy
to provide efficient dense raw numeric arrays, not to provide each
particular brand of array that anyone might want.

In your particular case what you need to do is write a small Python
class that wraps a numeric array and provides the extra functionality
you need for missing data.  Internally it would hold two arrays, one of
the actual data, and another one of ones and zeros indicating which data
was valid (just like Aaron suggested).

If you want to use NumPy on arrays of missing data you will need to
write this Python class yourself.  The point of NumPy is to make sure
that people have the efficient underlying array objects that they need
to build these higher level abstractions.  If you don't have the
time/inclination to build this small wrapper class yourself, then NumPy
is the wrong package for you.  If you do, perhaps you could contribute
your code back to the community so that more people can benefit from it.

-Jim

_______________
MATRIX-SIG  - SIG on Matrix Math for Python

send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________