I have a sequence prediction problem that confuses me. I am trying to predict some graphs as an output. I am using LSTM for this. I tried several things but still there is not any good results. I have 158 laboratory test which has 4 time series as an input and one time series as an output. I am taking 126 of them as training and 32 for testing. Input shape can be thought as (126,17658,4). Which are 126 samples, 17658 time steps and 4 features. Output shape is (126,17658) or it can be (126,17658,1). So, I am trying to predict a serie with looking to 4 time series. All have same timesteps. I tried classical LSTM structure, encoder decoder structure, LSTM with windowing etc. And there is not any accurate solution for this. What I am asking that if you faced with this problem, what will be the best method for this? Thanks.
Asked
Active
Viewed 33 times
