These two questions answer are very similar, and each has a good, highly upvoted, accepted answer: How to choose a predictive model after k-fold cross-validation?, and Training with the full dataset after cross-validation?.
It's also been pointed in comments to both questions that they are similar, and suggested they'd been merged somehow to keep the answers (I don't know if it's possible), or just mark one as a duplicate of the other (but protecting it from deletion).
The topic might be basic but it is active - I've marked two questions as duplicates while looking around for it. I marked them as duplicates of the first one (the answer seemed more straight-forward to me...), but it will be simpler to deal with (even for new users) if there's a single target for marking duplicates.
canonical-answertag. We do have a
canonical-questiontag, though, if it is one, but many of our merged questions don't have it. $\endgroup$
canonical-questiontag to use as an example, I don't really think I can judge that. $\endgroup$