Multi-Dataset Row Ordering

Multi-dataset row ordering uses the least common denominator approach for inter-sorting and any additional attributes for intra-sorting.

An example of row ordering.

Table A-1 Row Ordering

Dataset	SortAttributes
DS1	Employee,Account, Trans Am
DS2	Employee,Account, Logon ID
DS3	Employee,Account

In this example, the INTER sort occurs across datasets DS1, DS2, and DS3 using Employee and Account. Additionally, DS1 is INTRA-sorted on Trans Amt and DS2 is INTRA sorted on Logon ID.

Multi-dataset row ordering uses the minimum number of attributes listed for sorting for each dataset. The sort algorithm does not recognize names as are specified in the front end. Therefore, if you specify Employee as the first attribute in DS1 and Account as the first attribute in DS2 and DS3, the algorithm reads one attribute as the least common denominator and attempt to inter-sort on Employee and Account, which is incorrect. As data miner, you must specify the order of sort attributes for each dataset correctly so that the system performs the inter-sort on the same logical attribute. It is not necessary for an inter-sort attribute to have the same name across datasets, but they must have the same meaning and data type.