Discretization   theoretical foundation

The front end.
Post Reply
rafael.albuquerque
Posts: 2
Joined: Mon Jun 12, 2017 6:35 pm

Discretization   theoretical foundation

Post by rafael.albuquerque »

I would like to know if there is a theoretical basis for the method of uniformity. I'm waiting.
marek [BayesFusion]
Site Admin
Posts: 430
Joined: Tue Dec 11, 2007 4:24 pm

Re: Discretization   theoretical foundation

Post by marek [BayesFusion] »

You mean if there is any theoretical argument for optimality of uniform width or uniform counts? I am not aware of any general argument, although I would expect balancing the number of points in each of the intervals to perform better than having some intervals with few data points (unless there are good reasons for doing so but then you would want/need a supervised discretization method). I have been using the two as starting points for manual tweaking of the discretization bounds. I assume you know you can modify the bounds both by changing their numerical values in the table and by dragging them in the histogram plot (both in the discretization dialog).
I hope this helps.

Marek
Post Reply