This topic similar with this one R Time Series Analysis forecast result always remains same But I perfrom exponential smoothing model in R.
df=structure(list(date = structure(1:12, .Label = c("01.01.2017",
"01.02.2017", "01.03.2017", "01.04.2017", "01.05.2017", "01.06.2017",
"01.07.2017", "01.08.2017", "01.09.2017", "01.10.2017", "01.11.2017",
"01.12.2017"), class = "factor"), value = c(200L, 100L, 460L,
300L, 280L, 280L, 140L, 180L, 80L, 80L, 180L, 60L), common = structure(c(1L,
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), .Label = "158150- 01195т", class = "factor")), .Names = c("date",
"value", "common"), class = "data.frame", row.names = c(NA, -12L
))
i perform forecast for each groups separately) using ets function from forecast package (common is grouping variable)
here forecast for initial dataset
it is wrong result, but the main is that forecast on 1 year age ahead has static values(like historical mean)
How can i understand why forecast show same value in exponential smoothing model. Is it possible to fix it?
I can provide my R code, if it is needed.
ets()probably selects a non-seasonal and non-trended model, which yields flat line forecasts. Here are a few similar questions and answers (from me). – Stephan Kolassa Jan 10 '19 at 13:17ets()to use an additive seasonal model with the parametermodel="ZZA", or a multiplicative one usingmodel="ZZM". Unfortunately, there is no way to specify either one. However, ifets()chooses a non-seasonal model, it will likely have good reasons for it. I would be skeptical about being able to improve on its model selection. Do seasonal models truly improve out-of-sample forecast accuracy? – Stephan Kolassa Jan 10 '19 at 15:35ets()will not do anything seasonal. Why Does ETS Say My Data Has No Seasonality? – Stephan Kolassa Jan 10 '19 at 15:37