Software

Alfredo Roccato (Italy) has written an efficient SAS macro, %NBC, for a naive bayes classifier. It only requires SAS Base. This macro allows equal frequency or equal width methods for the discretization of continuous input variables and supports Laplace and M smoothing estimators. Optionally, it generates SAS data step code. The %NBC macro can be used in SAS Enterprise Miner within a SAS code tool: all output variables are fully compatible. Interested users can contact the author for the code via Alfredo.Roccato@unicredit.it .

Alfredo Roccato is a senior database marketing specialist for UniCredit Banca in Milan, Italy. Previously he has worked for SAS Institute Italy for nearly 13 years in various capacities including technical support, training, sales support and consulting, before becoming data mining product manager. He has been a SAS user since 1985.