We use the gganimate package in R. You can see the code used in these slides here: github.com/robjhyndman/fpp3_slides/blob/76096ba2ef1718149b71a3ff3ac9a887d97a3055/8-1-ses.Rmd#L66
Shouldn't the alpha=0 (orange) line be changing over time as we would be averaging everything in the previous steps? In this video, it is a constant line, which assumes that we know the mean of all the steps ahead of time (including the future steps we haven't seen yet).
The estimation of the model uses the entire training set, and what is shown is the fitted values within the training set. i.e., these are not real forecasts.
Thank you! Nice explanation!
Thank you for that video. Is there a possibility to set the initial value for l[0] manually?
No, we have not provided that facility in the software
4:39 how do you develop an animation? is there some R function?
We use the gganimate package in R. You can see the code used in these slides here: github.com/robjhyndman/fpp3_slides/blob/76096ba2ef1718149b71a3ff3ac9a887d97a3055/8-1-ses.Rmd#L66
Shouldn't the alpha=0 (orange) line be changing over time as we would be averaging everything in the previous steps? In this video, it is a constant line, which assumes that we know the mean of all the steps ahead of time (including the future steps we haven't seen yet).
The estimation of the model uses the entire training set, and what is shown is the fitted values within the training set. i.e., these are not real forecasts.