From the course: Python for Time Series Forecasting
Join today to access over 24,800 courses taught by industry experts.
Diagnostics to validate assumptions and inform model choice - Python Tutorial
From the course: Python for Time Series Forecasting
Diagnostics to validate assumptions and inform model choice
- [Instructor] The decision whether to pick SARIMA or exponentially smoothing depends on your goal and the application of the model. If you're going to forecast the next year, the resources, you better use the SARIMA model, because it's better specified on the assumptions, as I will explain later. And you can see that the forecast is basically the same than the exponentially smoothing, if you will go for a shorter term. But since we are having 96 months ahead, we can see that there is an exponential explosion on the SARIMA model for the long term, whereas this doesn't happen for the exponential smoothing. So if you're planning too many months ahead, I will stick with the exponential smoothing. Regarding the diagnostics and assumptions of this model, the exponential smoothing, we have the auto correlation plot where the significant values in the short term with 10 lags is not significant to the autocorrelation of the residuals, which is great, but it is on the medium- and long-term…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.