Marko Milicic
2008-Jun-16 21:48 UTC
[R] Constrained Optimized Binning Procedure....implementation help/idea needed.
Dear R Helpers, At the moment I'm working on the project to implement "optimal binning" function. It will be primarily used as a tool for logistic regression..... something very similar to http://www2.sas.com/proceedings/forum2008/153-2008.pdf* *but applied in diferent problem space...* *The problem might be descibed as finding optimal binning which will satisfy all of the rules imposed: - minimum and/or maximum percentage of observations per bin (good, bad, and total) - minimum and/or maximum bin width - minimum and/or maximum number of bins - minimum aggregate WOE difference between bins - monotonicity of aggregate bin WOE values (increasing, decreasing, or a combination) - etc... The idea is to use MILP (Mixed integer linear programming) and to solve problem..... Any ideas how to implement this would be welcomed... even better... if someone is aware of any package which allready has this functionaly... please us me know Thanks for help. -- This e-mail and any files transmitted with it are confid...{{dropped:14}}
Possibly Parallel Threads
- Binning function in R
- Binning of integers with hist() function odd results (P (PR#14047)
- Binning of integers with hist() function odd results (P (PR#14048)
- High density scatter plot with logarithmic binning
- crosstabs and histograms with flexible binning of dates