This site uses different types of cookies, including analytics and functional cookies (its own and from other sites). To change your cookie settings or find out more, click here. If you continue browsing our website, you accept these cookies.
@nathan_fredette, when covariates are used you need to use the TS Covariate Forecast tool rather than the TS Forecast tool. When using this tool, you will need to provide the needed covariates as well for the periods you want to forecast. If these are time lagged values of variables or variables that have a prori known values (such as the day of the week New Year's Day falls on), then doing this is straightforward. However, if they are current values (i.e., occuring at the same point as time as the variable you wish to predict that aren't known a priori), then you will need to provide forecasted values of the covariates. What this is means is that the prediction intervals provided will understate the true prediction intervals since they are based on the forecasted values of the covariates, which themselves contain error.
The TS Compare tool currently only works with univariate time series models. Altering this macro to work with ARIMA models that contain covariates should be fairly straightforward, and will included in a future release.
Error: TS Compare (67): Tool #18: Tool #6: No valid fields were selected. Error: TS Compare (67): Tool #19: Tool #6: No valid fields were selected. Error: TS Compare (67): TS_Compare: The time series models do not all forecast the same data field.
I am predicting hourly sales using Arima and ETS, same target variable and same input sample. please help