Population Dynamics Insights (PDI) provides a rich, location-specific snapshot of how populations engage with their surroundings. It captures this complex relationship in an embeddings dataset that can be integrated into your existing geospatial and predictive models.
PDI embeddings are generated by Google's proprietary model trained on a geo-indexed dataset of rich aggregated information from Google Search trends, and Google Maps' busyness visitation patterns, points of interest, weather and air quality data. Delivered as a 330-dimensional embeddings dataset in BigQuery, PDI enables organizations to understand the "DNA" of any location, fill critical data gaps, and model human-environment interactions globally with unprecedented precision and zero manual feature engineering. These embeddings ensure privacy while enabling nuanced spatial analysis and prediction for applications ranging from public health to socioeconomic modeling.
About PDI embeddings
Population dynamics embeddings are generated using a purpose-built machine learning model, trained on a rich set of features and converted into a condensed vector representation. These embeddings are trained on:
- Aggregated Search Trends: Captures the most popular and representative search patterns for a location to inform human behavior and interests.
- Aggregated Maps Point of Interest: Captures the types of businesses, establishments and landmarks present.
- Aggregated Busyness: Captures foot traffic visitation patterns to indicate density and frequency of human presence.
- Aggregated Weather and Air Quality: Captures the weather and air quality statistics and patterns.
These features are aggregated at S2 cell level 12 to generate localized, context-aware embeddings that preserve privacy.