BIOL 335 Lecture Notes - Lecture 17: Exponential Smoothing, Long Term Ecological Research Network
Document Summary
This activity aims to help you appreciate how the parameter in exponential smoothing can a ect the tted values and therefore the forecasts: consider the following gure, which shows precipitation levels over a course of 58 months. In addition to the data, tted exponential smoothing models with = 0. 05, = 0. 3, and. The case with = 0. 9 puts much weight on the last value, so the predicted series follows the original closely but looks as though it lags behind one time unit. As is usual, the smallest value of gives the smoothest model giving a good prediction of the overall trend : consider the following gure, which shows the stock price for a company over 14 days with. The same exponential smoothing models are t to this series as in question 1. Now here the lowest value of 0. 05 is not dynamic enough to capture the change in level apparent following the seventh value.