How the Home Environment Index is calculated
Exact inputs, weights, missing-data rules, quality gates and limitations for the current release.
Current AQI is shown separately and has 0% weight.
Why the method changed
Version 1 mixed a momentary AQI snapshot with slow-moving climate, housing and regulatory records. It also allowed unavailable values to behave like real zeroes. Version 2 separates current conditions from structural context and publishes no score unless every weighted component is available.
Component normalization
State drinking-water context · 35%
The score equals the percentage of active community water systems in a state that are not returned by EPA ECHO’s current health-based-violation filter. This remains a coarse state context signal and is labeled accordingly.
Climate moisture · 30%
Warm-season pressure = 55% normalized May–September mean temperature + 45% normalized May–September precipitation. Component score = 100 − pressure. Values are clamped to 0–100.
Housing age · 35%
Component score = 100 − the ACS share of housing units built before 1980. This is a maintenance and age-related-hazard context proxy, not a direct condition measure.
Missing-data policy
Scores are not renormalized around missing fields. If water, climate or housing is unavailable, the city’s score and rank are withheld. Current AQI may be unavailable without invalidating the structural index because it is not weighted.
Build-blocking quality gates
- At least 75 unique city records.
- Explicit available, unavailable or stale status for every component.
- AQI between 0 and 500 and traceable to AirNow.
- Water violation counts between zero and total active systems.
- No recurrence of the nationwide total-equals-violation filter failure.
- Twelve complete NOAA months and plausible annual precipitation.
- ACS percentages between 0% and 100%.
- At least 70 complete weighted city records before ranking publication.
Versioning
Every validated release is copied to a dated archive in the data directory. Method changes increment the methodology version. Live city URLs show the newest release; download records expose the version and generation timestamp.
What the index cannot tell you
It cannot determine indoor air quality, drinking water at a tap, mold, lead, radon, flood risk or the condition of a building. It is an editorial synthesis of public geographic data, and different weights would produce different rankings.