I've got a project I'd like to start working on and it roughly goes like this: I have past time series data for thousands of products that a company produces. There is a high weekly and yearly seasonality to all of the series and clear periods of increasing/ decreasing consumption across all series, though the products each have their own lifespan and consumption rates. There is also essentially a dial that can be turned for each of the products (i.e. a number between 0 and 1 that can be manually set). However, turning that dial has a cost associated with it so the problem is to analyze the past series data (along with how that dial has been turned for those products in the past) and predict how turning the dial will affect each product in the coming months. Essentially, what's the best bang for my buck?
I only know a bit of data science and not much at all about time series forecasting in particular. I've been researching models for the past week and my head is spinning with trying to understand all of the different ways people do time series forecasting (LSTM, GRU, VARIMA, VAR, NNAR, ETS, VECM, ... the list goes on and on).
I'm willing to spend some time learning, but I don't need/ want to become an expert in all of these methods before I can decide which one to use for my particular problem. I'm hoping you guys can help me narrow down which of these methods will work well for the particular problem I've taken on. Thanks.
P.S. I'd prefer something I can code up in Python since I already know that language.