When running a Boosted Tree Classification (or Random Forest Classification), considerably large data sets are displaying an error message 'Invalid Number Entered'. See example Booted Tree Workspace with error message below:

All

This error message is displayed in larger data sets and is due to the fact that number of records in a data set  affects the 'Min n% of cases' calculation. The default value for 'Min n % of cases' is 5%, and this value is too high for large data sets in Statistica 13.5. See options below for Boosted Trees, showing default 'Min n % of cases' is default set to 5:



To illustrate as an example, suppose you had a large data set with 4.5 million records. If the 'Min n % of cases' is left default at 5%, this would mean the min n of cases would be 225,000, which is too large, due to a limitation that the total (not percent) should be less than or equal to 100,000 observations in Statistica version 13.5 or lower. 

To overcome this error message and run the analysis, change the default 'Min n % of cases' from 5%, to a value less than 5 (until the total value for min n of cases is less than 100,000)

The limit of 100,000 observations occurs in Statistica version 13.5 or lower, and has been fixed in Statistica version 14.0 and onwards.

Error 'Invalid Number Entered' when running a Boosted Tree