I am trying to solve the problem of forecasting time series. I tried with ANN and LSTM, played a lot with different parameters, but all I could get was 8% better than the forecast for the duration.
So, I was wondering: since you can save models in keras; Is there any pre-prepared model (LSTM, RNN or any other ANN) for time series forecasting? If so, how do I get them? Does Keras have it?
I mean, it would be very useful if there was a site containing pre-trained models so that people don’t have too much time to train them.
Similarly, another question:
Can the following be done? 1. Suppose I now have a dataset and use it to train my model. Suppose that in a month I will have access to another data set (corresponding to the same data or similar data, in the future, perhaps, but not exclusively). Is it possible to continue training the model? This is not the same as training it in batches. When you do this in batches, you have all the data in one moment. Is it possible? And How?
source
share