Your data is here [![enter image description here][1]][1] . This is a cumulative series which will have false autocorrleation due to the accumulation.


 The correct thing to do is to analyze the observed daily series which is here [![enter image description here][2]][2] which has a level shift , five pulses and an an model (1,0,0) based upon an acf of [![enter image description here][3]][3] . 

The Actual/Fit and Forecast for the observed series is here[![enter image description here][4]][4] with forecasts here [![enter image description here][5]][5]

Now to forecast the cumulative simply take your last value and sequentially add each forecasted value.

The cumulative series is a derived series ( by addition) and will not provide the clarity that you need you need to analyze the observed daily series with whatever software that you have access to that identifies the arima model in the presence of a level shift and pulses.

What you are doing is a typical mistake of transforming the observations i.e.filtering them and then trying to create a model for the artificially summed series. 

Hope this helps you and others .


Build an ARIMA model like this one [![enter image description here][6]][6]


  [1]: https://i.sstatic.net/dP060.png
  [2]: https://i.sstatic.net/W8pH6.png
  [3]: https://i.sstatic.net/lrqRa.png
  [4]: https://i.sstatic.net/r5VN0.png
  [5]: https://i.sstatic.net/Ys6to.png
  [6]: https://i.sstatic.net/SzSWJ.png