I used RF and SGB(stochastic gradient boosting) statistical models for model prediction but I am getting only variable importance/Relative influence/InMSE list. How to get proper model equation from these significant variables?Whether I have to build model manually using these variables or the regression methods (RF, SGB) itself can give out model equation like stepwise multiple linear regression?
With a random forest model, there are dozens of models (forests) you generate. So, the splits for any given tree in the forest will be different.Some models will use certain factors. Other models will not.
The additive model can be suitable by considering the significant variables only for study. The model shall also depend on the nature of the variables like nominal, ordinal, interval etc. The suitability of logistic regression can also be checked (0/1 type variables)
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling