Papers for

real estate analysts

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

CartoVec encodes map-based location context without images

CartoVec: General-Purpose Location Embeddings from Open Map Data

Abstract: Geographic coordinates are common inputs in applied machine learning, but latitude and longitude describe position rather than the surrounding environment. Image-grounded location encoders learn this context from satellite or ground-level imagery; structured open map data offers another source of supervision whose utility for continuous coordinate encoders remains less explored. We introduce CartoVec, a coordinate encoder trained to predict Overture-derived land cover, land use, transportation, place and connectivity targets at multiple spatial scales, without image inputs or downstream-task labels. We evaluate frozen embeddings on Census outcomes, Airbnb prices and Philadelphia crime in the US Northeast using spatial block cross-validation. CartoVec has the highest mean performance on most Census outcomes and the highest average across the full suite, with several comparisons close across folds. The findings support open map data as a practical source of reusable location representations with a coordinate-to-vector interface that requires no map lookup at inference. The open-weight model is available at https://huggingface.co/gursoyfurkan/CartoVec.

Mon 12 OctMachine Learning
The gist
Geographic coordinates like latitude and longitude tell you where a place is, but not what the place is like. The authors created CartoVec, a tool that turns coordinates into detailed digital descriptions based on open map data, without using images or specific task labels. Their tool was tested on data about demographics, housing prices, and crime rates, where it often performed better than existing methods. CartoVec’s embeddings can be reused easily and do not require consulting maps when used.
Open → 2610.17190v1

Foundation model embeddings improve urban livability predictions

Applying foundation model embeddings towards urban livability evaluation

Abstract: While accurate measurement of socioeconomic indicators remains challenging in data-scarce regions, which limits policy interventions and resource allocation, high-resolution geospatial data is widely available and can contain information on various livability statistics. We investigate which physical features are encoded within foundation model embeddings, such as AlphaEarth, AnySat, and TerraMind, and provide a systematic framework for identifying the most predictive geospatial indicators. By analyzing how different types of geospatial data influence urban livability predictions, our approach enables researchers to prioritize the most informative features for their specific applications. Additionally, we demonstrate how to leverage foundation model embeddings to enhance prediction performance for these outcomes. This work contributes a principled methodology for extracting actionable information from satellite imagery while accounting for complex spatial dependencies, with applications in predicting urban livability in regions with limited observation data.

Tue 8 SeptMachine LearningComputers and Society
The gist
Measuring how livable a city is can be hard in places without much data. The authors look at special AI tools called foundation model embeddings that analyze satellite images to find clues about cities’ livability. They found a way to pick the best types of satellite data to make these predictions more accurate. This helps people understand which city features matter most for making places better to live, especially where traditional data is missing.
Open → 2609.09429v1