statistic on python (Rpy)

root root at sirius.ftsm.ukm.my
Mon Apr 10 11:41:03 CEST 2000


hai,

/* I am not an expert on C or python, but know enough to make my way around C
 * programs. This is very preliminary, and is primary to generate interest.
 */
 
currently there is not much support for doing statistical computing on
python (distributions, random numbers ). as an experiment, I took R src
code, (http://www.r-project.org),(R is  a computer language for doing
statistic , based heavily on S and Scheme), hack a bit and use SWIG to
implement the basic statistical distributions as python modules.

after a few days, I finally able to build a python module consist of basic
distributions. (the distributions computing is wrote in C)

my opinion is that, I would nice to port R statistical engine into python
module, and the user could use python as the front-end for statistical
programming. anybody willing to try this? this would greatly complement
Numpy, and help establish python as a computing platform.

R also have graphical interface.

Some interesting notes;

1. the basic distributions computation is in C, this can be imported into
      Python as module. however, the C src is a bit difficult to follow. I
      have to resort to try-and-error to finally get the module done.

2. the graphic is nice, not sure how port this to python.

3. some function is based on lapack. use in linear aljebra.

4. there quite a lot of R specific code that you have to consider in making
     the module. this is hard an time consuming.

5. most statistical function is writing in S, it is probably possible to do
     R2Py. Some of them are straight forward, and can be reimplemented using
     Numpy.

zamri




More information about the Python-list mailing list