Time series forecast. I have a model, how can I apply it to predict?


I have data from 2007 up to May 2015 and I would like to predict what is going to happend in the 2016.

I have already used, for the data up to May 2015, the lag column node combined with the linear regression learner.  What I do not understand is, how can I apply what I have learned to the new data?

I have no numbers to give to the lag column node, and I cannot link missing values to the regression predictor node. I encounted the same problem with missing values if I use the ARIMA nodes. Can anyone help me?

Thank you very much,



The data look like this:

2014-06-01 8030
2014-07-01 4813
2014-08-01 4850
2014-09-01 4839
2014-10-01 6570
2014-11-01 3466
2014-12-01 2477
2015-01-01 5053
2015-02-01 3433
2015-03-01 6099
2015-04-01 4728
2015-05-01 3463
2015-06-01 ?
2015-07-01 ?
2015-08-01 ?


Hi Lulu,

I tried and here's a workflow that should works (attached)

1) Exclude missing values for the learner 

2) replace missing values for the predictor (fix value = 0) before applying the prediction.

Hope that helps,