I was reading that ARIMA models are used to model the mean whereas GARCH models are usually used to model the conditional variance (i.e. volatility). Is it possible that the GARCH model can somehow be modified to predict the future value of the time series itself instead of the future volatility?
E.g. if you have historical data on the total amount of rain falls every week - could a GARCH model be used to predict how much rain will fall next week? Or is this impossible?