Recently I am trying to learn how to build multivariate LSTM models according to the guidance in Brownlee(2018).
I have data for 185 athletes with a total time length of 38 gameweeks, each athlete has 9 input features X, and a score variable Y. After 38 gw I expect to get a score forecast of 39,40 or more for each player. But I had some confusion when trying to build lstm.
Q1. There is no correlation between each player, is lstm a good choice?
Q2. In Brownlee 2018 it is stated that X(number of samples, number of time steps per sample, number of variables). So my input should be (185,38,9) and expect to get Y(185,1)?
If possible, can I get a code demonstration of a similar relevant case? Thank you very much!