0
$\begingroup$

I have time series data about a museum visitors in each month since 2011 to 2019. This data has seasonality. If I want to forecast the visitors in 2020, what forecasting method I should use? ARIMA, Simple Moving Average, or Exponential Smoothing? and why? Thank you :)

$\endgroup$
2
$\begingroup$

The Simple Moving Average and a non-seasonal ARIMA model will not capture seasonality, so if the seasonal signal is strong enough, try Exponential Smoothing with a seasonal component or seasonal ARIMA. (If the seasonal signal is weak, modeling it may make matters worse.) Take a look at the ets() and the auto.arima() functions in the forecast package for R. I strongly recommend Forecasting: Principles and Practice (2nd ed.) by Athanasopoulos & Hyndman.

That said, if you don't include the effects of the COVID-19 lockdowns in 2020, your forecasts will be way off. Whether and how to include these depends on what information set you have (i.e., should you assume that the lockdowns were already on the horizon at whatever date you calculated your forecasts?).

$\endgroup$
2
  • $\begingroup$ How do we include various effects of covid? Should dummy variables do just fine? My approach was to include '1's since pandemic declaration. $\endgroup$ Sep 23 '21 at 17:50
  • 1
    $\begingroup$ @AmanMathur: take a look at this thread. $\endgroup$ Sep 24 '21 at 9:45

Not the answer you're looking for? Browse other questions tagged or ask your own question.