Here it goes my solution
Prepped the data to use only data >= 2017. Set up the months, built the plot to check if everything was right.
Set up the ARIMA model and used the TS Forecast tool (did it for 96 months cause the targeting number was pretty high). Although if you check the plots you see that the model basically loses efficiency after 2020/2021, and it becomes linear.
Used the Upper boundary of the confidence interval (high). Unioned the forecasted values with the actual values and just did a running total.
Good way to exercise the time series tools, loved it. Cheers,
Interestingly, the forecast changes to 2026 if we forecast the total number of posts instead of the monthly posts.
On the community homepage, it says we are at 346.000 posts. This is closer to the 'total number' forecast, which predicted this number for March this year.
