Forecasting 'incomplete' periods

[Nikoleta-v3]

Forecasting has been used to forecast the net 10 years of the fields (thus the model is there).
One can argue though that the years 2014-2017 are missing information because the journals
that we are actually using have not made all of their articles available through the API system. Things like that take time.

Possibly, use the model to forecast each of this years but going back one time step each time. What
does the analysis show? Is that decreasing behaviour in the last years due to the incomplete information we are receiving?

[Nikoleta-v3]

Plot all these as well.