Features

The previous KDnuggets Poll asked:
How frequently did you use PMML (Predictive Model Markup Language) to deploy your data mining models in 2008

The results revealed a sharp division.

While about 70% replied never, 15% used it for more than 50% of the models.

US respondents (about half of respondents) showed the same pattern: 71% never used PMML, and 14% used PMML for more than 50% of the models.

This topic also generated a large number of comments.

Phil Brierley (Tiberius Data Mining) commented that PMML is too inflexible and said that their current solution is to generate scoring code in as many conceivable formats as possible such as SAS, SPSS, SQL, Excel, C etc. so most people can score data in their favourite application.

Alex Guazzelli (Zementis) commented that t is not very productive to replicate a single solution in many different formats (for different packages) if you could represent it in a single format, and that is why their company uses PMML.

Tim Manns opined that SQL is better than PMML for large databases

Here are all comments and full results of KDnuggets Poll: Using PMML to deploy data mining (July 2008)