DBbun Davis Square Synthetic Dataset (1800–2200)
A fully synthetic, privacy-free, and educational dataset simulating the evolution of the Davis Square area in Somerville, Massachusetts from the 1800s through the 2200s.
This dataset enables learners, researchers, and developers to explore data science, analytics, and machine learning safely — no real people, addresses, or businesses are represented.
Dataset Summary
| Table | Description |
|---|---|
geo_streets.csv |
Real street names with synthetic geometry |
geo_parks.csv |
Real park names used as location anchors |
poi_generic.csv |
Generic points of interest (restaurants, cafés, groceries, pharmacies, etc.) |
households.csv |
Synthetic household attributes (dwelling type, occupants, income, tenure) |
pets_registry.csv, pet_incidents.csv |
Pet ownership and neighborhood pet events |
mobility_trips.csv |
Trips by mode (walk, bike, car, bus, train) across eras |
public_safety.csv |
Synthetic safety incidents with categories and severity |
events_civic.csv |
Civic and festival events (e.g., HONK-style parades) |
observations.csv |
Environmental measures (noise, air quality, temperature, foot traffic) |
transit_* |
Transit lines, stops, and daily ridership |
bike_infra.csv, traffic_counts.csv |
Bicycle and vehicle infrastructure statistics |
prices_index.csv |
Long-term indices for housing, food, and transit fares |
weather_daily.csv |
Daily synthetic weather from 1900 to 2200 |
trees_inventory.csv |
Street-tree registry with species and heights |
infrastructure_events.csv |
Tree falls, water-main breaks, potholes, outages |
building_issues.csv |
Household maintenance and system failures |
DATA_DICTIONARY.json |
Column-level descriptions |
README.txt |
Summary of generated dataset |
All coordinates are synthetic and jittered for privacy; no real parcels, residents, or businesses appear.
Available Sizes
| Size | Approx. Households | Approx. Trips | Use Case |
|---|---|---|---|
tiny |
300 | 3 000 | quick demos, teaching syntax |
small |
2 000 | 25 000 | classroom exercises |
medium |
8 000 | 120 000 | research, ML prototypes |
large |
30 000 | 500 000 | performance and scaling |
xlarge |
80 000 | 2 000 000 | large-scale simulations |
Example Use
Two notebooks are included:
1. DBbun_Davis_medium_demo.ipynb
Descriptive analytics and visualization:
- Street and park mapping
- Household composition and mode share
- Noise, air quality, and event trends
- Weather and infrastructure summaries
2. DBbun_Davis_ML_demo.ipynb
Machine-learning examples:
- Classification: predict high-severity safety incidents
- Regression: predict noise levels (dB) from weather and traffic
- Clustering: K-Means grouping of streets by mobility patterns
Each notebook saves all figures, tables, and models locally.
Applications
- Teaching data wrangling, visualization, and machine learning
- Practicing geospatial analysis without privacy concerns
- Designing urban data dashboards and visual storytelling
- Benchmarking synthetic-data pipelines or evaluation metrics
- Running hackathons and bootcamps with realistic yet safe datasets
Privacy Statement
All data are synthetically generated.
Street and park names are used only as contextual anchors; all attributes, events, and metrics are algorithmically created with randomized geometry and time evolution.
No personal, identifiable, or proprietary information exists in this dataset.
Citation
Kartoun, U. (2025). Davis Square Synthetic Dataset (1800–2200) — A fully synthetic multi-era urban dataset for geospatial, mobility, environmental, and civic-event modeling. DBbun LLC.
- Downloads last month
- 23