...

I was going to suggest the same thing: type casting can have non-obvious effects, so explicit is better than implicit in this case.

Aye. Did you mean to send this just to me and not the list?

nope -- I know i"m in the minority but i really wish lists had reply-to set to the list. Bringing it back on. A pandas DataFrame comes with a conversion facility as a distinct

method, but it only copes with some basic types (int, float), and most of the DataFrame operations expect homogenous data. In fact, I think a given column is always homongenous except for columns of type "object", but you can't do math operations with "object" columns - they tend to be labels and stuff like that.

Sure -- Pandas is built on top of numpy arrays -- which have always been homogenous types -- which is quite deliberate -- numpy arrays provide three things: 1) N-dimensionsl arrays -- very handy 2) fast vectorized computations 3) a nifty Python wrapper around C (or Fortran) arrays. 2 and 3 are only possible with homogeneous (fundamental) types. The statistics module doesn't have the same limitations. But no matter how you slide it, if the types are not homogenous, then there will be some loss of some sort. Which is why I think it's better to make it explicit. It's actually hard for me to imagine a case where someone would have mixed types that wasn't an accident -- another reason to be explicit. -CHB -- Christopher Barker, PhD (Chris) Python Language Consulting - Teaching - Scientific Software Development - Desktop GUI and Web Development - wxPython, numpy, scipy, Cython