Skip to main content
SafeNSound

Transparency & Methodology

Full transparency about how SafeNSound works β€” every data source, every weight, every limitation.

Data Sources

15 sources across crime, socioeconomic, and environmental categories.

Crime

FBI NIBRS Crime Data

Sex offenses and kidnapping/abduction crimes aggregated by state

2023
Coverage: All 50 states + DCType: Official government data

Limitation: Aggregated by state only; not all agencies report to NIBRS

Sex Offender Registry (Aggregated)

Estimated registered sex offenders per 100,000 residents by state

2023–2025 estimates
Coverage: 50 statesType: Public registry estimates

Limitation: Estimates only; actual numbers may differ; aggregated at state level

Florida Sex Offender Registry

Individual registrant records aggregated to county level

Current
Coverage: Florida only (90,820 records β†’ 2,050 county-level aggregations)Type: Official state registry

Limitation: Only covers Florida; no individual data is displayed in the application

Socioeconomic

U.S. Census β€” Poverty Data

Percentage of families below the poverty line by state

Most recent Census
Coverage: 52 states/territoriesType: Official government data

Limitation: Poverty is a correlative indicator, not a direct safety measure

NCES School Data

Middle and high schools with location coordinates

Current
Coverage: 44,891 schools nationwideType: Official government data

Limitation: Does not include elementary schools or private institutions not in NCES

ICE Detention Centers

Locations of ICE detention facilities and documented protest sites

2023
Coverage: 49 locations (sample)Type: Public facility data

Limitation: Sample data; not a comprehensive list of all facilities

Retail & Commercial Locations (OpenStreetMap)

National inventory of retail chains: Walmart, GameStop, Hot Topic, Claire's, Sephora, Guitar Center, Baby Gap, Spencer's, McDonald's, vape/tobacco shops, shopping malls, and arcades sourced via Overpass API

2024–2025
Coverage: 15,000+ locations nationwide across 13 retail categoriesType: Community-maintained open data

Limitation: OSM coverage varies; some locations may be missing or have outdated data

Youth & Family Organizations (OpenStreetMap)

Boy Scouts, Boys & Girls Clubs chapters, daycares/childcare centers, and community centers sourced via Overpass API

2024–2025
Coverage: ~7,700 daycares, 826 Boy Scout units, 58 Boys & Girls Clubs chapters, 3,000+ community centersType: Community-maintained open data

Limitation: OSM coverage varies by region; private/unlisted organizations not included

Law Enforcement & Higher Education (OpenStreetMap)

Police stations/precincts and college/university campuses sourced via Overpass API

2024–2025
Coverage: Nationwide via OSMType: Community-maintained open data

Limitation: Smaller/rural departments and community colleges may have lower OSM coverage

Scientology Organizations (OpenStreetMap)

Church of Scientology locations sourced via Overpass API

2024–2025
Coverage: ~6 confirmed OSM-tagged locations (sparse coverage)Type: Community-maintained open data

Limitation: Very low OSM coverage β€” many locations not tagged; data is incomplete

Parks & Recreation (OpenStreetMap)

Parks, playgrounds, and recreational spaces sourced via Overpass API

2024–2025
Coverage: Nationwide via OSMType: Community-maintained open data

Limitation: OSM coverage varies by region; rural areas may be under-mapped

Environmental

NOAA Storm Events Database

Severe weather events (tornadoes, floods, hurricanes, hail, high winds, etc.) aggregated by county with severity weighting

2019–2024
Coverage: All US countiesType: Official government data

Limitation: Aggregated county-level event counts only; weighting is approximate and does not capture sub-county variation

FEMA National Flood Hazard Layer (NFHL)

FEMA Special Flood Hazard Area (SFHA) coverage percentage by county and state, derived from official flood maps

Current FEMA mapping
Coverage: Continental USType: Official government data

Limitation: Based on available FEMA flood maps; some areas lack detailed surveys and may be under- or over-estimated

EPA Toxic Release Inventory (TRI)

EPA-tracked industrial and federal facilities that report toxic chemical releases, aggregated by county

2023
Coverage: 32,755 facilities nationwideType: Official government data

Limitation: Covers only facilities required or choosing to report under EPCRA Section 313; does not capture all pollution sources

Sample Data

Missing Persons Reports

Missing persons and kidnapping cases by county

2023
Coverage: ~73 sample countiesType: Sample / Mock data

Limitation: THIS IS SAMPLE DATA for demonstration. Not actual missing persons statistics.

Data Freshness

All data is static as of the dates below. SafeNSound scores reflect a snapshot β€” not a live feed. Crime patterns, registries, and facility inventories change over time. ↑ = updated to latest available 2025 source.

SourceCoversAs of
FBI NIBRS Crime DataViolent crime by county2023 ↑
NSOPW Sex Offender RegistryRegistered offenders by ZIPRolling (live)
FBI Missing Persons (NIBRS)Missing persons by county2023 ↑
US Census ACSPoverty rate by county2023 5-yr ↑
NCES School Locations44,000+ school coordinates2024–25 ↑
OSM / Google PlacesParks, daycares, playgrounds2025
ICE Detention Facilities49 confirmed facility locations2025
NOAA Storm EventsSevere weather density 2020–20252025 ↑
FEMA NFHL Flood MapsSpecial Flood Hazard Area coverage2025
EPA TRI FacilitiesIndustrial/chemical facility density2024 ↑
HUD Fair Market Rents2BR FMR by county (affordability)FY 2025 ↑
EIA Electricity PricesResidential Β’/kWh by state2025 Jan–Apr ↑
EIA Gasoline PricesRetail $/gal by state2025 Jan–May ↑
EIA Home HeatingOil & gas heating costs by state2024–25 season ↑
USDA Food Access AtlasLow food access share by county2023 ↑
BLS Consumer ExpenditureFood & lifestyle costs by state2024 ↑
AWWA Water RatesResidential water & sewer $/mo2025 survey ↑
IHRSA Gym MembershipAverage gym cost by state2025 survey ↑

How Scores Are Calculated

A weighted composite of eight normalized factors. Higher score = more risk.

Factor Weights

Violent Crime Rate
27%
Sex Offender Density
23%
Kidnapping & Crimes Against Children
14%
Poverty Rate
9%
Child Safety Proximity
9%
Detention & Protest Sites
5%
Commercial Density
5%
Environmental Hazards
10%

Normalization

Each factor is min-max normalized across all states before weighting:

normalized = (value βˆ’ min) Γ· (max βˆ’ min) Γ— 100
total_score = Ξ£ (factor_score Γ— factor_weight)

Environmental Hazard Composite

The Environmental Hazard factor (10% weight) is itself a composite of three sub-signals, equally blended at the state level:

  • β€ΊStorm Frequency β€” NOAA severe weather event density, 2019–2024
  • β€ΊFlood Risk β€” FEMA SFHA coverage as % of county area
  • β€ΊIndustrial Footprint β€” EPA TRI facility density per county

On the hex map, Storm History, Flood Risk, and Industrial Footprint are also independently toggleable filter layers at local resolution.

What the Scores Mean

0–20Very Low RiskLowest measured risk factors in available data.
21–40Low RiskBelow-average risk compared to national baseline.
41–60ModerateRisk factors near the national average.
61–80ElevatedAbove-average risk across multiple categories.
81–100High RiskHighest measured risk factors in available data.

Important Disclaimers

β€’SafeNSound scores are statistical estimates for educational purposes only, not guarantees of safety or predictions of danger.

β€’A low score does not mean an area is unsafe; a high score does not mean imminent danger exists.

β€’This tool does not identify, target, or track any individuals. Sex offender density scores are aggregated at the county/state level.

β€’Offender map pins represent registered sex offender counts aggregated by ZIP code β€” no individual names, photographs, or exact addresses are displayed.

β€’Missing persons scores use FBI NIBRS kidnapping/abduction data normalized per 100k residents. County-level centroid pins are approximate geographic representations of county-level aggregates, not precise incident locations.

β€’The time-of-day modifier is a general heuristic based on criminological research β€” not actual time-specific crime data for any location.

β€’Environmental hazard scores (storm frequency, flood risk, industrial footprint) reflect historical and regulatory data, not current conditions or real-time alerts.

β€’This tool is not a substitute for personal judgment, local knowledge, or law enforcement advice.

β€’SafeNSound does not collect, store, or share any user location data.

Known Limitations

Geographic coverage gaps: Offender map pins use ZIP-code centroid coordinates for all states. Florida has the highest fidelity (individual registry records aggregated to ZIP). All state density scores use aggregate per-capita estimates.

Data coverage gaps: ICE detention site data covers a 49-location sample of confirmed facilities. OSM-sourced retail and POI data may have gaps in rural regions. Missing persons use FBI NIBRS aggregate data β€” county pin locations are centroids, not incident coordinates.

Environmental data resolution: NOAA storm and FEMA flood data are aggregated at the county level. EPA TRI covers only reporting facilities β€” not all industrial pollution sources. Sub-county variation is not captured.

No real-time data: All data is static as of the dates indicated. Crime patterns, flood maps, and facility inventories change over time.

Correlation vs. causation: Poverty rate is included as a correlative risk factor based on research. It does not imply that poverty causes crime or that impoverished areas are inherently dangerous.

Scoring model limitations: The weighting system reflects general research priorities but is not validated by an expert panel. Different weightings would produce different scores.

Bias & Fairness

Any data-driven scoring system can reflect and potentially amplify existing biases in the underlying data.

Crime statistics may be influenced by reporting practices, enforcement patterns, and systemic factors that disproportionately affect certain communities. A higher score does not necessarily mean an area is more dangerous β€” it may reflect differences in data collection and reporting.

Environmental hazard layers reflect regulatory tracking and historical weather records. Industrial facility placement has historically correlated with socioeconomic disparities; we present this data as informational context, not as a judgment about any community.

SafeNSound is designed to inform, not to label. We do not target any individuals, groups, or protected classes. All insights are derived from publicly available aggregate datasets.

Sex Offender Registry Coverage by State

Public sex offender registries are known to be incomplete. SafeNSound applies correction factors derived from DOJ compliance audit studies and independent state-level assessments to account for registrants who are non-compliant with address verification. The table below shows the estimated percentage of registrants NOT verifiably at their listed address.

StateEst. Non-CompliantCorrection AppliedRegistry Grade
NY37%Γ—1.59D
NJ31%Γ—1.45D
IL30%Γ—1.43D
MA28%Γ—1.39D
DC28%Γ—1.39D
MD26%Γ—1.35C
RI26%Γ—1.35C
CT24%Γ—1.31C
MT24%Γ—1.31C
ME23%Γ—1.30C
HI25%Γ—1.33C
NM25%Γ—1.33C
PA25%Γ—1.33C
VT25%Γ—1.33C
AK22%Γ—1.28C
DE22%Γ—1.28C
MS22%Γ—1.28C
NH22%Γ—1.28C
WV22%Γ—1.28C
WY23%Γ—1.30C
AR20%Γ—1.25C
IN20%Γ—1.25B
KY20%Γ—1.25B
MO19%Γ—1.23B
NE20%Γ—1.25B
UT19%Γ—1.23B
CO19%Γ—1.23B
ID19%Γ—1.23B
KS18%Γ—1.22B
NC18%Γ—1.22B
SC18%Γ—1.22B
TN18%Γ—1.22B
AZ18%Γ—1.22B
GA18%Γ—1.22B
AL15%Γ—1.18B
FL15%Γ—1.18B
TX16%Γ—1.19B
CA14%Γ—1.16A
MI17%Γ—1.20B
NV16%Γ—1.19B
WA17%Γ—1.20B
LA17%Γ—1.20B

Source: DOJ SMART Office compliance audit estimates, state-level independent assessments. Grade A = <15% non-compliant; B = 15–21%; C = 22–26%; D = >26%. Correction factor multiplied by registry density before scoring.

Why This Platform Exists

The public safety intelligence market is dominated by tools built for governments and institutions β€” not individuals. The companies that control this data charge tens of thousands of dollars per year, operate behind closed APIs, and have recently faced serious trust issues. SafeNSound was built to close that gap.

How We Compare

LexisNexis Accurint Risk Suite

2024 data breach exposing 364k+ individuals. FTC settlement for selling vehicle telematics to insurers without disclosure. Enterprise pricing starts at $23,000+/year.

100% public federal data. No PII handled. Free tier available. Full methodology published here.

Palantir Gotham

$1B+ DHS contract, $30M+ ICE ImmigrationOS (2025). Methodology classified. No public access. No consumer product.

Built for the public, not for governments. Every formula on this page. No government contracts.

SoundThinking (ex-ShotSpotter)

Persistent false-positive accuracy debates. Chicago non-renewal. NYC audits raised concerns. $104M annual revenue from municipal contracts.

No sensors. No acoustic surveillance. Only publicly available federal datasets anyone can verify.

Citizen / Ring Neighbors

Citizen: ~$192M funded, AI transcription accuracy disputed, radio scanner with privacy concerns. Ring: Amazon device flywheel, past law enforcement data sharing controversies.

Complementary, not competitive. Citizen and Ring show what just happened. We show what the data says over time. Use both.

πŸ“‚

Open Sources

Every dataset is publicly available and linked. FBI NIBRS, Census Bureau, NOAA, FEMA, EPA. You can download our source data yourself.

βš–οΈ

Published Weights

Every factor weight is shown on this page. If you disagree with a weight, you can say so β€” and we'll read it. No black boxes.

πŸ”’

No Surveillance

We do not operate sensors, cameras, or microphones. We do not purchase private data. We do not resell your location or behavior.

SafeNSound has no government contracts, no law enforcement data sharing agreements, and no institutional investors with access to user data. This platform is funded by optional subscriptions from users who want deeper features β€” not by selling data about the people who use it.

Radiation & Nuclear Risk Monitoring

Coming Soon

SafeNSound will surface publicly available Emergency Planning Zone (EPZ) data from the U.S. Nuclear Regulatory Commission. EPZs are not hazard scores β€” they are federally designated planning boundaries used by local emergency management agencies. Being inside an EPZ does not indicate danger; it indicates that the area is covered by an emergency preparedness plan in the event of a nuclear facility incident.

DatasetSourceCoverageUpdate Frequency
10-Mile EPZ BoundariesNRC PRIS / public GeoJSONAll 58 operating US reactorsQuarterly
50-Mile Ingestion ZoneNRC guidance / derivedAll 58 operating US reactorsQuarterly

Data ingestion in progress. EPZ boundaries are a matter of public record and are published by the NRC under 10 CFR 50.47. This layer will not display radiation levels or real-time incident data β€” only the static planning zone boundaries.