Creating a causal time series definition

1. Right click in the Causal Time Series Definitions pane and select New Time Series Definition, or click the icon at the top of the pane.

   

2. Enter a Name and a Description for the definition, then specify the input data source and options for the causal time series fields.
3. Choose your Input options -
   • Database Connection - Select the database containing the table to be aggregated, or choose New Database Connection to establish a new connection to another database.
   • Input Table Name – Select the table containing the causal time series data.
   • Time Stamp Field - Select the field containing valid time data.
   • Source - Select the type of source data being used to create the causal time series.
     • User Defined - causal data originating from user-defined tables imported into Data Guru; used to measure effects of various causal factors on demand
     • Promotion - promotional data originating from user-defined tables imported into Data Guru; used to measure the effect of promotions on demand (data must consist of boolean values)
     • Trend Cloud - filtered collections of time-based causative data provided through Data Guru's Trend Cloud Import action; used to explore external factors that drive customer demand
4. In Time Series Fields, select those fields from the input table that will be used to create the causal time series.
   • Click the Add icon to add a causal field to the time series. You can add multiple causal fields.
   • For each causal field you add, choose the Field containing the causal data, and select the Aggregation method for the causal field.

     Aggregation methods available when the Source is either User Defined or Trend Cloud include -

     • Average
     • Min (default)
     • Max
     • Sum

     Aggregation methods available when the Source is identified as Promotion allow you to determine what happens when promotion information is available a more granular level than the time bucket to be used for forecasting. Choices include -

     • If Any (default) - A promotion is applied during a particular time span if there is at least one period within the time span that contains the promotion.
     • If All - A promotion is applied during a particular time span only if all the time periods within the time span contain the promotion.