Which dataset I should use when I am retraining my model?

Question

I trained my deep learning model using x dataset and now I got new dataset and I give it name as y. I want to retrain my model on this new dataset which is y. Do I need to use x+y dataset or just y? Can you tell me what is right approach and what are the effects of that?

If nothing like that and other approaches are available please post it that helps a lot.

Thanks in advance.

Answer 1

So, I understand the question is asking what data should you use, when given new data. You might want to provide more context to your question as this might seem like a loose answer. The way you go forward with this depends on a couple of factors:

1. Is the initial dataset ($D_1$) have the same features/columns as the new dataset ($D_2$), such that you can integrate them (as you said). This obviously can help to answer whether to use either $D_1$/$D_2$ or both.
2. It also depends on the problem specification: if this new dataset provides new features (maybe more accurately measured), which are more appropriate to the problem spec, then obviously you should then work with $D_2$ and train the model with this dataset. If the problem spec can be solved using both datasets, then use both as this can help to improve generalisation performance more so than using $D_2$ alone.