Tabular learning, multiple weighted training sets?

Hi, I’m relatively new to deep learning and wondering if there’s a good method for training for my current use case:

I want to predict whether a building is going to sell in the next 12 months from a list of buildings. I have a building dataset and a list of sales dataset going back 20+ years. While I can just filter for the last years worth of sales and train it on that, I was wondering if there was a better method that could also account for earlier years’ data and maybe just weigh it differently (there might be some insights in there). I’m not sure how to go about doing this with fastai so any hints/tips/suggestions would be helpful.