Automatic feature generation

fungayism

hie. 
How do you find or calculate the minimum number of iterations of N attributes exampleset using Backward Elimination.

Telcontar120

Backward elimination will run one model using all attributes and that is the "baseline" model performance.  Then in the next round, it will run a model dropping each one of the attributes individually and check the performance of each one.  At that point, if none of the reduced models are acceptable (based on whatever threshold criteria you have set in the Backward Elimination operator parameters) then it will terminate.  Thus the minimum number of iterations using Backward Elimination is (n+1) where n is the number of available attributes.

Telcontar120

Backward elimination will run one model using all attributes and that is the "baseline" model performance.  Then in the next round, it will run a model dropping each one of the attributes individually and check the performance of each one.  At that point, if none of the reduced models are acceptable (based on whatever threshold criteria you have set in the Backward Elimination operator parameters) then it will terminate.  Thus the minimum number of iterations using Backward Elimination is (n+1) where n is the number of available attributes.

fungayism

thanks.
 i was confusing it with n+(n-1)