Chris Fonnesbeck wrote:
On 4/3/06, *Matthew Vernon* <matthew@sel.cam.ac.uk <mailto:matthew@sel.cam.ac.uk>> wrote:
My concern is that this is effort re-inventing a wheel. Would it not be better to improve the R interface rather than effectively duplicating functionality that's already in R (a Free, and high quality piece of software)?
I strongly disagree with this approach. I'm not interested in dropping back to a 3rd party application to do sophisticated statistical analysis. We should be able to do it in python, and do it well. I specifically chose python to implement my Bayesian analysis software because it is faster, more flexible, and more user-friendly than R. In fact, all of scipy could have been replicated in R if desired, but that would be a bad choice.
I'm of this opinion as well. Certainly we can learn from R (especially regarding interfaces and useful functions to implement). -Travis