Hello everyone,
Can anyone help?
I know that if our independent variables and response variable are time
series then our model has ARIMA errors.
Now, does anyone have any references containing good, clear examples
where the response variable is again a time series but the predictors
contain categorical variables?
For example, say our time series was 100 hours long and we had a usual
time series response variable Y( recorded every hour) and a usual 'time
series' type predictor variable X (recorded every hour)and a categorical
predictor variable in our model,'Activity', with Activity A being
carried out during hours 1 to 10, Activity B being carried out from
hours 11 to 40, Activity C being carried out from hour 41 to 78 and
Activity D being carried out from hour 79 to 100. How would this
variable be incorporated in the model? As 3 dummy binary variables? If
so, what would happen if the response and predictors required
differencing?
Many thanks for your help,
Kim
|