Presenting decision tree
Posted by Oleg Solovyev on Mar 14, 2010
Presenting data analysis results to the business people is one of the hardest tasks. One has to avoid technical terms and explain math concepts in simple and clear language. Recently I faced this problem again having to present decision tree model to my director.
The best idea I came up with was to print the decision tree on A2 size paper. The idea worked well because the print attracted a lot of attention and provoked a long discussion. This article is devoted to the tricks to improve the decision tree schema that is created by the SAS Enterprise Miner Decision Tree node.
Maximum levels of 512
Posted by Oleg Solovyev on Feb 28, 2010
Working with Enterprise Miner I do often encounter an error “Maximum levels of 512 exceeded”. It occurs after adding new table to the project during the execution of one of the model nodes like decision tree or logistic regression. The reason is that one of the nominal variables has more than 512 different values.
Enterprise Miner vs SAS/STAT
Posted by Oleg Solovyev on Dec 15, 2009
From a conversation:
– Wait. So, where is Data Mining implemented?
– Look here (drawing). Initially they developed Base SAS then SAS/STAT and SAS/OR. These are libraries if math algorithms developed using C language. Later they developed Enterprise Miner using Java. EM is just a set of wizards that run procedures in SAS/STAT and SAS/OR. Knowing the procedures EM runs one can run those procedures without EM.
– In other words EM creates Base SAS code and runs it in SAS.
– Yes, that is right.
– So, do you refuse to use EM at all?
– Well, No. Sometimes I use EM to build decision trees manually.
– OK. Then what EM was developed for?
– Good question. One can also ask what EG was developed for?
– It is easy to develop reports in EG. Three clicks and it is ready.
– I can create report using Base SAS. Three lines of code and report is ready. The thing is that there are people that don’t like coding. I do like coding and I do it well. EM and EG were developed for those who don’t know about SAS/STAT or don’t like to code.
SAS products overview
Posted by Oleg Solovyev on Jun 6, 2009
Before discussing SAS software let’s consider two terms that SAS is actively using describing its software. The terms are “product” and “solution”. “Product” is the irreducible software that performs one particular function. For instance products are: DBMS’s, clients’ for DBMS, web portals, etc. “Solution” is a set of products that automates process in a particular business field. SAS for instance has solutions for banking, Telco, insurance, etc. As a rule, solutions contain predefined database schema, user interface and reports optimized for a particular task.
