NYC Urban Analytics Hub -- Community Collaboration Invitation

Hi Cesium Community,

I’ve started a project called NYC Urban Analytics Hub, which brings real population and urban datasets into Cesium for Unreal and applies predictive analytics to them. The aim is to allow simulation and exploration of “what-if” scenarios in a spatial context – for example, selecting a piece of land in New York City and projecting statistics or outcomes based on given parameters.

I’ve set up a Hugging Face collection where the project is being organized and updated:
:backhand_index_pointing_right: NYC Urban Analytics Hub Collection

This project is still in its early stages, and I’d love to build it further with the community. Contributions are welcome directly in the Hugging Face collection and/or in any repositories connected to this project. Areas for collaboration include:

  • Expanding datasets and pipelines

  • Enhancing predictive modeling (time series, TensorFlow, etc.)

  • Building Unreal Engine integrations and visualizations

If you’d like to be a contributor, feel free to DM me directly.

Looking forward to working together and seeing what we can build as a community!

Ali

Hi @alidenewade, welcome to the community! And thanks for sharing your project.

You might have to pardon my ignorance here, but where does Cesium for Unreal enter into your project? I explored the Hugging Face collection briefly, and couldn’t spot it.

Hi @Kevin_Ring, thanks for the warm welcome and for checking out my Hugging Face collection! :grinning_face:

Ah, I see. Gr8 question!

Ok. You’re right! HF repo itself mainly shows the models and data pipelines. Cesium for Unreal enters the project at the visualisation and simulation stage. After I generate outputs (for example, population insights, spatial risk models, or other geospatial/actuarial data), I bring those results into Cesium for Unreal to place them on a real-world 3D globe and create interactive simulations.

So, the HF part is focused on the AI/ML side (training, datasets, descriptors), while Cesium for Unreal is where I actually map and visualise those insights in 3D space. They complement each other: Hugging Face is the “engine” that processes the data, and Cesium for Unreal is the “stage” where I bring it to life for exploration and presentation.

I haven’t pushed the Unreal files to Hugging Face yet (since they’re heavier and better suited to a game engine repo, and would cost me an arm and a leg), but that’s the connection I had in mind.

I’m happy ro compile a quick UE5.6 project and link a Google Drive if you’d like. Let me know :grinning_face:

Ok, thanks for the explanation. No need to send me a Google Drive link, or add anything to Hugging Face on my account. But the audience here on the Cesium for Unreal community might be interested in how you’re going about visualizing this sort of data with Cesium for Unreal, if you’re willing to share.

Thanks, Kevin. That makes sense.

Yeah, sure. Ok. Let me give a real estate focused example of how I’m combining HF outputs with Cesium for Unreal. So, imagine a planner chooses a block in New York City and sets project parameters; things like number of residential units, retail space, lighting upgrades, or expected daily foot traffic. On the HF side, I run models trained on historical incidents, zoning, demographics, and environmental factors to predict crime rates over the next 12–24 months. The outputs come out as georeferenced grids or vector layers, showing baseline predictions and scenario adjustments based on those project inputs.

From there I bring those results into Cesium for Unreal, where I stream NYC’s 3D terrain and buildings as context. The predictions are layered directly on the map as heatmaps, extruded “crime bars,” or isolines that highlight high-risk areas. Planners then can adjust parameters with simple sliders (for example, adding more lighting or changing retail density) and immediately see how the predicted hotspots shift. Clicking a building or cell opens detailed stats like baseline vs. scenario rates and confidence intervals.

This workflow makes it easier to explore/assess “what-if” scenarios in real time. Instead of static charts or tables, decision-makers get a 3D, immersive view of how their design choices might change risk on the ground, helping them weigh whether to proceed with a project and which mitigation strategies to prioritise.

Right, so how did you do those things? Rendering vector layers or heatmaps in Cesium for Unreal is far from trivial, so it’s interesting to hear how you approached it. Or is this just a concept that you haven’t tried to implement yet?

Sure. So, I’m approaching this through Cesium for Unreal’s Edit Tileset Materials and Visualising Metadata features, and also tweaking native Unreal Engine Blueprints to drive dynamic effects like heatmaps, isolines, and simple extrusions. Still a work in progress, which is why I wanted to get the community’s input on more efficient ways to handle vector-like overlays in Cesium for Unreal.