Discussion: View Thread

Jack Ring forwarded me the thread below. 

Debate topic: Hard decisions are made by teams of people managing incomplete, uncertain, evolving and conflicting information. Any method not based on the support of uncertainty and not capable of rolling up new information as it becomes available is weak.

I am the author of  Accord.  See www.robustdecisions.com.  You can download a 30 day trial , but we will give a free license to academic institutions.  Also, see the book, Making Robust Decisions.

If you want more details or you want me in the "debate", please let me know.

541-754-3609

Hi Gary,

I play with R, the open source programmable math programming language. IN there I use a windows Gui called Rattle, also free. The guy that writes it has programmed in there for me a transformation called Nolans Groups and Matrix for me.

What my concept does is take several measurement variables, and then clusters them with an indexing calculation based on several descriptive variables. So each item being clustered receives a rating, which is relative  to its peers within that sub population. Once every item is indexed, then we apply a fuzzy matrix, and then a kmeans to the matrix which clusters into 1000 groups. Then using that kmeans result as a target variable, we then run a decision tree or random forest to make a set of rules for dissecting and modeling the data.

We are calling it called Nolans Relativity Peer Clustering Index.

So far we still havent broken it, we have tried to over load the matrix, we have changed random number seeds, we have independent entity rating systems, and then bench mark the clusters against our output and the clusters are true to their rules in the difference form the original populations norms and averages.

So what used to be complex analysis, now using the newer analytics tools is becoming childs play, for people to invent new approaches, all you need to do is forget the past faults of statistic and maths, realise we are in a fuzzy not crisp world, and model the interactions.

Regards
Tony Nolan




__________ Information from ESET Smart Security, version of virus signature database 4546 (20091027) __________

The message was checked by ESET Smart Security.

http://www.eset.com