I am using Statsmodel Holt Winter library, and am able to create a model successfully. Now I need to create a csv file which will have in one column actual values and in another predicted (along with suitable index).
Now for my testing model it is quite easy, I can just take my test dataset and use model.forecast()
from Holt Winter library but am not able to figure out how to get the prediction for training data.
My thinking is to take prediction from both training and testing data and display it alongside the original dataset.
For finding in place values i.e. predicted values of training data, there is a function called fittedvalues
in Holt Winter library.
mdl.fittedvalues
This will give us predicted values of our training data.