Automating time series forecasts

The Expert Modeler functionality in Modeler greatly simplifies time series forecasting. The Time Series node will automatically determine which model type is most appropriate for your data: ARIMA, exponential smoothing, seasonal model, and so on. However, in practice, a time series model nugget can only generate forecast models for a single time series. It is possible to generate multiple time series forecasts using the Time Series node but it is largely impractical. First, you must pivot the data such that each series is a column. Second, defining input variable roles can become convoluted due to each field having only a single role (for example, a field cannot be an input for one series but none for another input). Finally, you must reverse-pivot the forecast data back to the original format to make use of it. This reverse pivot requires you to have a fixed set of input names to pivot. With all of these limitations, the Time Series node does not scale to...