"Neural Networks for prediction using historical figures"

C00221919

Hi, 

 

Looking for some help guys, I have constructed a Neural Net in Rapidminer with historical numerical data, with the most recent year being the label.  I get no correct predictions, however each record or row is predicted to within about 10% which may be acceptable for this purpose. Can you explain please how:

 

1. Can I now apply the model to predict the figure for 2018, as the neural net operator will not accept a column with no data present.
2. Can I write the the data out to an excel file, when i introduce the "Write Excel" operator it only writes out the original example data set and not the new data set with the prediction included.

I would greatly appreciate some assistance on this matter as my process is part of my dissertation for my Masters Degree and I am working against the clock.

 

Thanks in advance, look forward to hearing from you.

 

Kindest Regards,

Claire Murphy

Thomas_Ott

Also search through the Community for Time Series and Windowing too.