Prediction with Deep Learning / LSTM

ToSchnieToSchnie Member Posts: 1 Newbie

So far, I have done time series forecasting (e.g., stock value forecasting) using autoregression ARIMA, using the operators Retrieve--Filter Examples--ARIMA--Apply Forecast.

Since the extensions for LSTM layers are now available, I wanted to try the time series forecasts with Deep Learning.

To do this, I simply replaced the ARIMA operator with the Deep Learning operator [Extension]. Similarly, I replaced Apply Forecast with Apply Model (Generic) [Extension].

For the inner structure of the Deep Learning operator I chose the following sequence:
Add Fully-Connected Layer with neurons = 1 and activation function = ReLU; Add LSTM Layer with neurons = 10 and activation function = ReLU; Add Output Layer with output type = Automatic.

Unfortunately, this simple model does not work and gives the error message "Network configuration problem".

I would be extremely grateful to the experts here if you could give me some helpful tips?

Regards Torsten
Sign In or Register to comment.