Joining Record Types Versus Linking Datasets
In SuiteAnalytics Workbook, joins within a dataset are realized through SQL queries that are executed in the background. Specifically, a left outer join is executed for each record type you add to a dataset. This means that the results include all the rows from the primary record type, plus any rows that match the join condition from the joined record type. No aggregation is performed, and some results may be overstated or understated depending on the relationship between the record types and the order that you join them in. For more information, see Guidelines for Joining Record Types in SuiteAnalytics Workbook.
By contrast, linked datasets behave more like an SQL union. In a union, two queries are executed to produce one set of results, with the union being defined by a set of fields that share the same data type within each query. In Workbook, this means that each dataset you link is executed as a query in the background, with the results being aggregated and displayed based on the fields you define as common keys when you create the link.
To summarize, joins in a dataset enable you to combine and analyze fields from multiple record types in a single set of results. Linked datasets enable you to analyze and aggregate the results of two queries as though they were executed as one query. For more information about linking datasets and common keys, see Link Datasets in a Workbook.
To learn more about the differences between joining record types in a dataset and linking datasets in a workbook, see the following table:
|
|
Joining Record Types |
Linking Dataset |
|
SQL execution |
Left outer join. Results include all primary record fields, plus matching joined record fields. |
Union. The results from two datasets are displayed as one visualization, with only distinct values included in the results and aggregation based on the fields you define as common keys when you create the link. For more information, see Common Keys. |
|
Number of data sources |
Unlimited. You can join any number of record types in a dataset, however too many joins can result in performance issues. |
Two datasets. |
|
Data duplication |
Data duplication is possible depending on the relationship between the record types you join, and the order in which you join them. For more information, see Data Duplication Based on Record Joins. |
No. Results only contain distinct values from each dataset, so there is no data duplication. |
|
Aggregation |
No. This is intentional, since datasets are meant to be used as raw data for your visualizations. |
Yes. Aggregation occurs automatically based on the fields you define as common keys and where you place them in the visualization. |
|
Sharing |
Yes. After you join record types in a dataset, you can save and share it with other users in your account. |
Yes, but only as part of the workbook they are linked in. |
|
Requirements |
|
|