Skip to main content

​
Overview

OpenAVM Kit can automatically enrich your property data with geographic features from OpenStreetMap (OSM). This adds distance-based features that measure proximity to important amenities like:
  • Water bodies (lakes, rivers, reservoirs)
  • Parks and green spaces
  • Educational institutions (universities, colleges)
  • Transportation infrastructure (major roads, railways)
  • Golf courses
These distance features can significantly improve property valuation models by capturing location-based amenities.

​
How It Works

OpenStreetMap enrichment:
  1. Downloads OSM data for your locality
  2. Filters features by type and size (e.g., parks larger than 2,000 m²)
  3. Identifies top features (e.g., the 5 largest parks)
  4. Calculates distances from each parcel to nearby features
  5. Adds distance fields to your dataset
No API key required! OpenStreetMap data is freely available under the Open Database License.

​
Configuration

Add OpenStreetMap enrichment to your settings.json: 
{
  "process": {
    "enrich": {
      "universe": {
        "openstreetmap": {
          "enabled": true,
          "water_bodies": {
            "enabled": true,
            "min_area": 10000,
            "top_n": 5,
            "sort_by": "area"
          },
          "parks": {
            "enabled": true,
            "min_area": 2000,
            "top_n": 5,
            "sort_by": "area"
          },
          "educational": {
            "enabled": true,
            "min_area": 1000,
            "top_n": 5,
            "sort_by": "area"
          },
          "transportation": {
            "enabled": true,
            "min_length": 1000,
            "top_n": 5,
            "sort_by": "length"
          },
          "golf_courses": {
            "enabled": true,
            "min_area": 10000,
            "top_n": 3,
            "sort_by": "area"
          }
        },
        "distances": [
          {
            "id": "water_bodies",
            "max_distance": 1500,
            "unit": "m"
          },
          {
            "id": "water_bodies_top",
            "field": "name",
            "max_distance": 1500,
            "unit": "m"
          },
          {
            "id": "parks",
            "max_distance": 800,
            "unit": "m"
          },
          {
            "id": "parks_top",
            "field": "name",
            "max_distance": 800,
            "unit": "m"
          }
        ]
      }
    }
  }
}

​
Feature Types

​
Water Bodies

Rivers, lakes, reservoirs, and other water features.
​
openstreetmap.water_bodies.enabled
boolean
default:"false"
Enable water body enrichment
​
openstreetmap.water_bodies.min_area
number
default:"10000"
Minimum area in square meters. Filters out small ponds and streams.Recommended values:
  • 10000 (1 hectare) - Significant water bodies only
  • 5000 (0.5 hectare) - Include medium-sized features
  • 1000 (0.1 hectare) - Include smaller features
​
openstreetmap.water_bodies.top_n
number
default:"5"
Number of largest water bodies to track individually. The top N features will have individual distance fields created.
​
openstreetmap.water_bodies.sort_by
string
default:"area"
Property to sort by when selecting top N features. Use area for water bodies.

​
Parks and Green Spaces

Public parks, gardens, playgrounds, and recreational areas.
​
openstreetmap.parks.enabled
boolean
default:"false"
Enable parks enrichment
​
openstreetmap.parks.min_area
number
default:"2000"
Minimum park area in square metersRecommended values:
  • 2000 (0.2 hectare) - Neighborhood parks and larger
  • 5000 (0.5 hectare) - Community parks and larger
  • 10000 (1 hectare) - Regional parks only
​
openstreetmap.parks.top_n
number
default:"5"
Number of largest parks to track individually

​
Educational Institutions

Universities, colleges, and other educational facilities.
​
openstreetmap.educational.enabled
boolean
default:"false"
Enable educational institution enrichment
​
openstreetmap.educational.min_area
number
default:"1000"
Minimum campus area in square metersRecommended values:
  • 1000 (0.1 hectare) - Small colleges and up
  • 5000 (0.5 hectare) - Medium campuses and up
  • 10000 (1 hectare) - Large universities only
​
openstreetmap.educational.top_n
number
default:"5"
Number of largest educational institutions to track individually

​
Transportation

Major roads, highways, railways, and transit infrastructure.
​
openstreetmap.transportation.enabled
boolean
default:"false"
Enable transportation enrichment
​
openstreetmap.transportation.min_length
number
default:"1000"
Minimum feature length in meters. Filters out small road segments.Recommended values:
  • 1000 (1 km) - Major routes only
  • 500 (0.5 km) - Include medium routes
  • 100 (100 m) - Include most routes
​
openstreetmap.transportation.top_n
number
default:"5"
Number of longest transportation routes to track individually
​
openstreetmap.transportation.sort_by
string
default:"length"
Property to sort by. Use length for linear features like roads and railways.

​
Golf Courses

Golf courses and related facilities.
​
openstreetmap.golf_courses.enabled
boolean
default:"false"
Enable golf course enrichment
​
openstreetmap.golf_courses.min_area
number
default:"10000"
Minimum golf course area in square metersRecommended values:
  • 10000 (1 hectare) - Small courses and up
  • 50000 (5 hectares) - Standard courses only
  • 100000 (10 hectares) - Large courses only
​
openstreetmap.golf_courses.top_n
number
default:"3"
Number of largest golf courses to track individually

​
Distance Calculations

The distances array defines how to calculate distance features: 
{
  "distances": [
    {
      "id": "water_bodies",
      "max_distance": 1500,
      "unit": "m"
    },
    {
      "id": "parks_top",
      "field": "name",
      "max_distance": 800,
      "unit": "m"
    }
  ]
}
​
distances[].id
string
required
Identifier for the feature typeAggregate distances:
  • water_bodies - Distance to any water body
  • parks - Distance to any park
  • educational - Distance to any educational institution
  • transportation - Distance to any transportation route
  • golf_courses - Distance to any golf course
Individual distances:
  • water_bodies_top - Distance to each named water body
  • parks_top - Distance to each named park
  • educational_top - Distance to each named institution
  • transportation_top - Distance to each named route
  • golf_courses_top - Distance to each named course
​
distances[].field
string
Field to use for naming individual features. Typically "name" for top N features. Omit this for aggregate distances.
​
distances[].max_distance
number
required
Maximum distance to calculate in the specified unit. Features beyond this distance will be marked as null or max distance.Recommended values:
  • 800 m - Walking distance (parks, schools)
  • 1500 m - Short drive (water, golf courses)
  • 3000 m - Medium drive (universities, major amenities)
​
distances[].unit
string
default:"m"
Unit of measurement. Currently "m" (meters) is standard.

​
Output Fields

OpenStreetMap enrichment creates distance fields in your dataset:

​
Aggregate Distance Fields

Distance to the nearest feature of each type:
  • dist_to_water_bodies_any - Distance to nearest water body (meters)
  • dist_to_parks_any - Distance to nearest park (meters)
  • dist_to_educational_any - Distance to nearest educational institution (meters)
  • dist_to_transportation_any - Distance to nearest transportation route (meters)
  • dist_to_golf_courses_any - Distance to nearest golf course (meters)

​
Individual Distance Fields

Distance to each of the top N named features:
  • dist_to_water_bodies_lake_travis - Distance to Lake Travis (meters)
  • dist_to_parks_zilker_park - Distance to Zilker Park (meters)
  • dist_to_educational_university_of_texas - Distance to University of Texas (meters)
  • dist_to_transportation_interstate_35 - Distance to Interstate 35 (meters)
  • dist_to_golf_courses_barton_creek - Distance to Barton Creek Golf Course (meters)
Field names are automatically sanitized (lowercased, spaces replaced with underscores) to ensure compatibility with most data formats.

​
Example Configurations

{
  "process": {
    "enrich": {
      "universe": {
        "openstreetmap": {
          "enabled": true,
          "parks": {
            "enabled": true,
            "min_area": 2000,
            "top_n": 5,
            "sort_by": "area"
          }
        },
        "distances": [
          {
            "id": "parks",
            "max_distance": 800,
            "unit": "m"
          }
        ]
      }
    }
  }
}

​
Using in Models

The distance fields are automatically added to your dataset and can be used in models: 
from openavmkit.pipeline import run_pipeline

# Run pipeline with OSM enrichment
run_pipeline(
    locality="us-tx-travis",
    settings_file="in/settings.json"
)

# The output dataset includes OSM distance fields
# These fields are automatically classified as "land" features
The OpenStreetMap distance fields are automatically recognized and classified:
  • Fields starting with dist_to_ are treated as numeric land features
  • They’re included in feature selection and modeling
  • No additional configuration needed

​
Performance Considerations

Start SmallBegin with a small number of feature types and increase as needed. Each feature type adds processing time and data size.
Use Appropriate ThresholdsSet min_area and min_length values to filter out insignificant features. This improves performance and model quality.
Limit Top NTracking too many individual features can create hundreds of fields. Start with top_n: 3-5 and increase only if needed.

​
Best Practices

​
Choose Relevant Features

Select feature types relevant to your market:
  • Residential: Parks, schools, water bodies
  • Luxury residential: Golf courses, water bodies, large parks
  • Urban: Transportation, parks, educational institutions
  • Suburban: Parks, schools, golf courses

​
Set Appropriate Distances

Consider typical travel modes in your area:
  • Walking distance: 400-800 meters (parks, schools)
  • Short drive: 1,000-2,000 meters (amenities)
  • Longer drive: 2,000-5,000 meters (special features)

​
Balance Granularity

Use both aggregate and individual distances:
  • Aggregate (parks) - Captures general proximity to amenity type
  • Individual (parks_top) - Captures proximity to specific high-value features

​
Troubleshooting

​
No Features Found

Issue: OpenStreetMap enrichment runs but no features are found. Solutions:
  • Lower the min_area or min_length thresholds
  • Verify OSM has data for your locality
  • Check that feature types exist in your area (e.g., golf courses may be rare in urban areas)

​
Too Many Features

Issue: Hundreds of distance fields are created. Solutions:
  • Reduce top_n to track fewer individual features
  • Increase min_area or min_length to filter smaller features
  • Focus on aggregate distances instead of individual features

​
Slow Processing

Issue: OpenStreetMap enrichment takes too long. Solutions:
  • Reduce the number of enabled feature types
  • Increase minimum size thresholds
  • Reduce top_n values
  • Process a smaller geographic area

​
Data Quality Notes

OpenStreetMap Data Varies by RegionOSM data quality and completeness varies significantly by location. Urban areas typically have better coverage than rural areas. Always verify the features make sense for your locality.
Feature Names May Be MissingNot all OSM features have names. Unnamed features will be skipped when creating individual distance fields. This is normal and expected.

​
Next Steps

Census API

Add demographic data enrichment

Settings Configuration

Learn more about settings.json structure

Data Processing

Understand the data processing pipeline

Feature Engineering

Learn about feature engineering techniques

Build docs developers (and LLMs) love

Get started for freeTalk to us