The input table must contain id columns for the time series and the period within the time series (the timepoints).
To ensure repeatability, the user has the option of creating a table using the CCMPrepare function. This function adds a column, “aster_ccm_id”, to the input table so that the partitioning of data across workers is guaranteed to be consistent over multiple function calls. If the InputTable contains an “aster_ccm_id” column, the function assumes that data has been prepared using CCMPrepare. If it does not contain this column, the function generates this partitioning column internally.
The Aster CCM function supports categorical variables as possible causes and effects. This feature is to be considered experimental.
Column Name | Data Type | Description |
---|---|---|
aster_ccm_id | Integer | Optional. Partition column used internally by the function to ensure repeatability across multiple function calls. If this column is not present, the function generates it automatically. |
sequence_id | Any except float or double | ID of the time sequence. |
timeperiod_id | Any | Time associated with each value in the timeseries values columns. |
timeseries_values | Any | Any number of columns. Values from the time series to be tested for potential causal relationships. |