Search code examples
amazon-web-servicesamazon-quicksight

Lost calculated fields

I am using Amazon Quicksight for some data analysis and visualisation.

I wanted to extend existing data with a new table in datasource view -> add data link. after joining this new table with existing one, data refresh and switching to data analysis i realised that i have lost all calculated fields i had in my project. So now i have following questions:

  1. Anyone else encountered similar issue?
  2. Is it possible to recover lost fields?
  3. is it possible to go back to a state before change was done? It seems that Undo option does not work for changes done in data sets
  4. how to protect my analysis against such unexpected behaviours in future?
Solution

  • Yes, I have experience this before.

    There are two types of calculated fields, one at the data set level and one at the analysis level (aggregated calculated fields).

    For the data set level fields, it is definitely recreate-able behavior that when you add a new table to join, all calculated fields are removed (as of the writing of this post at least). Once you click Save, there is no way to get them back that I am aware of. Best you can do it just document all of the calculated fields and re-add them after you add the join table.

    For aggregated calculated fields at the analysis level, those don't seem to be lost after you change the dataset. I have seen them get into a state where they are no longer working since some of the fields for the underlying dataset have changed so you need to edit them and correct the fields they reference.

    Some practices that I have learned to help avoid accidentally introducing unwanted changes:

    1. Duplicate you dataset. If you click on the data set in the data set view and there is a "Duplicate Data Set" button. Use this feature to play around with changes to a dataset if there are visualizations that are dependent on it.

    2. When working on an analysis, click "Autosave On" at the top-center to disable auto save. This allows you to make experimental changes and play around with the visualization accidentally saving some small tweaks.

    Hope that helps.