logistic regression

[pradyumna reddy]

Hello Sir, 

I am Pradyumna, I am coding a enthusiast, I code in C,C++ and Python , my subjects of interest are machine learning and game theory. I came across simpleai recently, Simple AI is really interesting project, from few days i am trying to include logistic regression classfiers into simpleai, the optimizing function which i want to include is Roland's python port of Carl Rasmussen's minimize.m , I want a advice on which is a better idea  either to include minimize function as a classmethod to logistic regression class in classifier module or to create a separate module in name of tools 

[Rafael Carrascosa]

Hi Pradyumna, I don't know much about logistic regresion, but my
recomendation is that the optimizing function should go with the
problem description, in a subclass of `ClassificationProblem`.

Perhaps you could add a `optimize_function` (or some other name)
abstract method to `ClassificationProblem`, and then create a new
class that inherits from `ClassificationProblem` and implements Carl
Rasmussen's minimize.

If Carl Rasmussen's minimize turns out to be a common way to deal with
these problems we could include it in simpleai too :)

Cheers!

> --
> Has recibido este mensaje porque estás suscrito al grupo "simpleai" de
> Grupos de Google.
> Para anular tu suscripción a este grupo, envía un correo electrónico a
> simpleai+u...@googlegroups.com
> Para obtener más opciones, visita https://groups.google.com/groups/opt_out.
>
>

[pradyumna reddy]

Thanks