search

facebook / prophet

Tool for producing high quality forecasts for time series data that has multiple seasonality with linear or non-linear growth.
https://facebook.github.io/prophet
MIT License
18.26k stars 4.51k forks source link

When building Prophet models should I consider differencing as a transformation to make my series stationary? #1029

Closed rydevera3 closed 5 years ago

rydevera3 commented 5 years ago

Hello all!

I have been doing a lot of work around time series forecasting as of late. When my series are not stationary I consider different methods to transform the series to be stationary one of the best is often n-levels of differencing which models like ARIMA already incorporate. When I do not have a stationary series is it beneficial to use differencing before modeling with Prophet?

daikonradish commented 5 years ago

Prophet isn’t ARIMA. In general you don’t need to have stationarity for Prophet to work well. See #234.