I wish to forecast electricity power demand at a mall. Data is collected daily, but, the data has multiple seasonality :

  1. data/power consumption is generally similar from monday to thursday but see a change on friday, saturday, sunday (weekly seasonality) ?

  2. power consumption vary with month. Summer months the power consumption is higher due to cooling required ?

  3. Year on year there is increase in power consumption with deployment of more equipments. (Yearly seasonality) ?

Series contain only date and power consumption (daily data). What seasonality should I use ?

Option 1:

powerts <- ts(data,start=2016,frequency=7) 
modArima <- auto.arima(powerts)

Option 2:

powerts <- ts(data,start=2016,frequency=365) 
modArima <- auto.arima(powerts)

But both seems lagging as they fail to cover all the three points above



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.