Bumped by Community user
    Tweeted twitter.com/StackStats/status/1120658729064632320
2 edited tags
| link
1
source | link

Forecasting daily data with annual seasonality

i have been trying to do the forecasting model. My data has daily value and there is annual seasonality and probably weekly. My question is which model will be the best. I have tried with SARiMA but i think it is not good enough for annual seasonality in this case(too long period) ? Maybe ARMA + fourier would be better ? The second case is that this data is strongly correlated with weather. Can i use for example SARiMAX with additional variable or VAR model? If someone has met with such data, I would be grateful for the hint which model will be good.

Below attached data chart

enter image description here

This is short example of code:

y <- msts(dane2$value, seasonal.periods=c(1,7,365.25))
fit <- tbats(y)
fc <- forecast(fit)
plot(fc)

enter image description here

I would like to get more accurate results. Every idea will be appreciate !