Dashboard choropleth map with geometry

[der]

We are currently building our dashboards in Apache Superset. With the Git integration in Databricks AI/BI Dashboards, the development process has improved a lot. So we are thinking about switch to Databricks AI/BI Dashboard.

One pain point in Databricks AI/BI Dashboards is the map visualization for custom choropleths. Unfortunately, this is not supported 😥, which means no custom regions can be displayed:
https://learn.microsoft.com/en-us/azure/databricks/dashboards/visualizations/maps#choropleth-options

In contrast, Apache Superset offers much more flexibility through deck.gl:
https://superset.datatest.ch/superset/dashboard/deck/

According following video by @MathieuDB  there is WIP for Geometries
https://youtu.be/Cx-FzcszqVI?t=1157

Is there a roadmap for this feature? @mjohns @dbkent

[NandiniN]

Hi @der ,

Fantastic choice to switch to Databricks AI/BI Dashboards! They are super awesome.

Unfortunately we do not have custom GeoJSON/choropleth support currently, but it’s a common feature request and internally tracked with DB-I-14646, DB-I-5257 and are considered for development. The roadmap does not specify a fixed timeline, but I have gone ahead and added a vote on this feature request by you, which will help to get them more attention.

Doc - https://docs.databricks.com/aws/en/dashboards/visualizations/maps#choropleth-options

Would you meanwhile, like to workaround by creating a mapping table where your custom region IDs are linked to the required built-in Databricks region level?

Thanks!

[der]

Hi @NandiniN ,

Thanks a lot for upvoting the feature requests. I really appreciate you pushing this forward!

Regarding the suggested workaround: unfortunately, a mapping table won’t solve our use case, since we need a much more granular map with custom regions.

From my perspective, this is becoming essential: Databricks is actively promoting the vision of bringing spatial data directly to analysts in SQL Warehouses. To fully realize this, analysts also need the ability to visualize the new geometry types from Unity Catalog natively in dashboards. Otherwise, the workflow breaks down. Analysts would have to switch at the very last step into a Databricks Notebook and use Python just for visualization, which undermines the goal of seamless spatial integration in SQL Warehouse.

Would be great if this gap could be prioritized on the roadmap, as it directly impacts adoption of spatial features.

[NandiniN]

Hi @der , 

I totally understand the challenge you are facing, not all use cases can get helped with the workarounds. I see you have tagged the authors of https://www.databricks.com/blog/introducing-spatial-sql-databricks-80-functions-high-performance-geo... in the original query. I will have a quick chat with them. 

In addition, I would also request you to have this discussion with your Account Executive (AE) and they can also request with the product team on prioritization of the feature.

TBH, there is a good adoption, and many customers are requesting for this feature, I do believe this should be out soon, but on ETA, I can update you after the internal discussion.

Thanks!