Let's say that I have a vector of features that I use to get a single result back using some machine learning algorithm.

I thought about using multiple variations of that algorithm to get multiple results back and then simply concatenating them together into one big features vector for a new algorithm to give me that one result back.

Would that enhance accuracy or make it worse?


This is widely used method, called stacking, see e.g. Bagging, boosting and stacking in machine learning

To answer your question: yes, it is very useful in some cases. If you cook it properly with folding, it almost sure to have quality no worse than a single model.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.