-2

While working in R, I came to know that

SES:-The simple exponential smoothing function, which takes into account random fluctuations of data but it does not take into account any trend or seasonality. Then Holt's model which takes into account of any of any trend or random data but not seasonality.

And the Holt's winter model takes into takes all the three scenario data like random, trend and seasonality.

So, I am bit confused everytime if we use Holt's winter method , it will always calculate all the scenarios of time series data i.e random, trend and seasonality? Although there is no randomness or trend. So, which family of smoothing data, shall I prefer to use always, modeling with timeseries data?

Please, anybody can explain it?

I hope my questions are clear. If not please comment me.

Avijit Mallick
  • 37
  • 1
  • 8

1 Answers1

0

R has ets package which applies exponential smoothing with trend and seasonality on data. But I would suggest using ensemble method of forecasting for accuracy. Ensemble method uses ForecastHybrid package for generating point forecast combing all advance algorithm (Holt Winter, ARIMA, Neural, STM, TBAT) at once.

Avijit Mallick
  • 37
  • 1
  • 8