I need to create a calculated field on a geospatial data type field in-Analysis. How do I make the in-analysis calculated field geospatial?
Hi @Ben_Berliner,
In your dataset you can create your calculated field and then change the datatype to geospatial. Here is some additional information on changing data types in QuickSight.
Hi @Peter thanks for your response. Unfortunately, it does not resolve my question. I am referring strictly to calculated fields created from within an Analysis. In this environment, there is no option to change the data type to geospatial.
Is there perhaps something I can do with calculated field code such as “CAST” or “CONVERT”?
Hi @Ben_Berliner currently there is no way to convert a calculated field created in an Analysis into a geographic field. You would need to do it in the Data Set itself.
I will mark your post as a feature request and our PM team will evaluate it for a future release.
Thanks @Jesse!
@Ben_Berliner what is your use case for only being able to create this calculated field in an analysis and not being able to create the same calculated field in the dataset? Is it an issue of you not having permission to edit the dataset?
@Peter , we’re working with a large dataset and would prefer not to need to add additional fields to it. Adding a new calculated field to the dataset requires the refreshing of the entire dataset, which is resource and time intensive.
This is a specific use case just for this particular Analysis so ideally we would much prefer to create the calculated field in-Analysis.
@Peter, as at late 2024, this feature has clearly not been implemented.
I agree with @Ben_Berliner 100% in that it is highly inconvenient to stack multiple calculated fields into a large dataset. Ideally, we’d want these as disposable analysis-only driven entities.
For example, I need to create a geospatial field to conform to QuickSight’s un-documented handling of States (in Australia), eg: NSW → “New South Wales”. I do not want this bloating my dataset, and just need it to fix QuickSight’s flaws, so I want this field to strictly remain in the analysis.
Additionally, having to add a calculated field directly to the dataset is highly inconvenient for the following reasons:
- I cannot rapidly iterate the development of the calculated field as a geospatial field, so I cannot immediately see the results of my testing, and,
- I need to again load the data into SPICE, thus spinning up the supporting cluster to feed the data in, which is not convenient.