Data Science Stack Exchange is a question and answer site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field. It only takes a minute to sign up.
Sign up to join this community
I am building an insurance recommendation engine. I have used some variables, like demographics, and built the model. Now I have claims data.
Is there a way to include the new data without restarting the process like this?
model1 = initial variables
model2 = new variables
model3 = x*initial variables+y*new variables
You can build a model two with new variables and a third one which is trained using predictions of first two models. This is called stacking. But in your case I would try to build a new model with all variables. Basically this is because you loose (not use) the information from the interaction between old set and new set of variables, which is often very useful. Usually stacking is used to combine different approaches not to split input space.
As far I know, if you want to extend your model training, then number of feature should be same as previous one.
In case you have more features, then you have two options:
- Drop those new features
- Retrain your model from scratch
In case you have less features, then again you have two options:
- Place some values in missing features. Yo can use Impute classes, Refer Link for more information
- Retrain your model from scratch
