Output unexplicable about seriesnet HOT 5 CLOSED

I think the reason because you don't run a lot of epochs or you are doing a wrong shape.
but the problem is when trying to feed the network with the time series is making it as input size of the network as in code take the length of the time series and passing it to DC_CNN_Model function did you face it ?
I have run the code in sequence input for 2000epochs and get reasonable results

timeseries = np.array([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16])
predict_size = 3
predictions = evaluate_timeseries(timeseries, predict_size)

but my problem as i see the input shape of the model will be the size of the time series.

from seriesnet.

I have the same issue as @cbouyssi . Just using the script as is and the example time series from @MohamedLotfyElrefai . The network is not learning and prediction is [0,0,0]. Maybe it is due to different package versions? keras==2.2.4 , tensorflow==1.12.0, numpy==1.16.2 or do I have to adjust anything?

from seriesnet.

I have the same issue mentioned by @fmmix .

from seriesnet.

I think the reason because you don't run a lot of epochs or you are doing a wrong shape.
but the problem is when trying to feed the network with the time series is making it as input size of the network as in code take the length of the time series and passing it to DC_CNN_Model function did you face it ?
I have run the code in sequence input for 2000epochs and get reasonable results

timeseries = np.array([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16])
predict_size = 3
predictions = evaluate_timeseries(timeseries, predict_size)

but my problem as i see the input shape of the model will be the size of the time series.

Hi @MohamedLotfyElrefai:
I run your example. And I got the same issue as others.

Could you please have a look? Thank you.

from seriesnet.

see updated script to see if that helps

from seriesnet.