[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Enhnancements to octaveforge/statistics/table
From: |
forkandwait |
Subject: |
Re: Enhnancements to octaveforge/statistics/table |
Date: |
Sun, 7 Feb 2010 05:30:37 +0000 (UTC) |
User-agent: |
Loom/3.14 (http://gmane.org/) |
> It is actually part of Octave core. So you should check out the Octave
> source code and not the Octave-Forge code.
As I contemplate patching this, I would like to propose an interface change,
such that if one passes in a n x 2 vector, assume that that vector should be
calculated with weights, while if it is n x 1, it is calculated without
weights. Otherwise throw an error.
If I was writing the API from scratch I would do the following: It would seem
more graceful to have table take an N x M matrix of data points for the first
parameter, with each of the M columns a different variable. The optional
weights should be a matrix of the same dimension. Then as the function is
extended to more dimensions than 2, we don't add parameters. I am just
mentioning this for discussion purposes; I don't want to break api's for this.
Thoughts?