Amazon QuickSight
User Guide

Changing a Data Set

There are two situations where you should be concerned about changes to a data set. One is if you deliberately edit the data set. The other is if your data source has changed so much that it affects the analyses based on it.

Analyses that are in production usage should be protected so they continue to function correctly. We recommend the following when you're dealing with data changes:

  • A best practice is to carefully document your data sources, data sets, and which visuals rely upon them.

  • When you edit a data set try not to make changes that will break existing visuals. For example, don't remove columns that are being used in a visual. If you must remove the column, create a calculated column in its place. The replacement column should have the same name and data type as the original.

  • If your data source or data set is changed in your source database, you will need to adapt your visual to accommodate the change, as described previously. Or you can try to adapt the source database. For example, you could create a view of the source table (document). Then if the table changes, you can adjust the view to include or exclude columns (attributes), change data types, fill null values, etc. Or, in another circumstance, if your data set is based on a slow SQL query, you could create a table to hold the results of the query.

  • If the data source is no longer available to you, your analyses will be empty. The visuals you created still exist, but they cannot display until they have some data to show. You can edit the visual and select a new data set to use with it. Make sure it has a similar structure to the original.