Note that we don’t include time i = 1 in the calculations of MAE and MSE.īy simple algebra, this iteration can also be expressed asĪnd similarly for all values of ŷ i. In particular, for some α where 0 ≤ α ≤ 1, for all i > 1, we define In Simple (aka Single) Exponential Smoothing, the forecasted value at time i+1 is based on the value at time i, and the forecasted value at time i (and so indirectly on all the previous time values). Exponential Smoothing improves on Weighted Moving Average by taking all previous observations into account, while still favoring the most recent observations. In Weighted Moving Average, you can give more weight to recent events, but you are limited to the last m observations.
0 Comments
Leave a Reply. |