> ## Documentation Index
> Fetch the complete documentation index at: https://docs.risk.matrisk.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Data Overview

> Overview of the data we collect and how it is used

### Coverage

* **85% of total premiums** written across the U.S. 100M+ records.
* **Vehicle-level data** available for several states.
* New data will be added as we expand to additional states.

### Timeframe

* Depending on the state, **5 – 20 years** of historical crash data.
* Continuous updates to capture changes over time.

### Data Refreshing

* **Lag:** New data arrives with a **1 – 12 month** delay.
* Refresh frequency varies by state and data source.

***

### Geographic Coverage

{" "}

<img className="block dark:hidden max-w-sm rounded-lg" src="https://mintcdn.com/three-sigma-data/krO97y3WKndm6k1i/images/crash-coverage.webp?fit=max&auto=format&n=krO97y3WKndm6k1i&q=85&s=6949d3046b022184ba001e00969ef849" alt="Crash Coverage Light" width="867" height="498" data-path="images/crash-coverage.webp" />

<img className="hidden dark:block max-w-sm rounded-lg" src="https://mintcdn.com/three-sigma-data/krO97y3WKndm6k1i/images/crash-coverage-dark.webp?fit=max&auto=format&n=krO97y3WKndm6k1i&q=85&s=a10f2ab625d34af3b8e0bea5511c09f4" alt="Crash Coverage Dark" width="841" height="493" data-path="images/crash-coverage-dark.webp" />

* **Finished:** Green states are fully onboarded with historical data. - **Launching
  Q2 2025:** Peach states will be available soon.{" "}

***

## Crash Data Lifecycle

The lifecycle of crash data involves multiple steps to ensure completeness, accuracy, and standardization.

{" "}

<img className="block max-w-sm rounded-lg" src="https://mintcdn.com/three-sigma-data/krO97y3WKndm6k1i/images/crash-lifecycle.webp?fit=max&auto=format&n=krO97y3WKndm6k1i&q=85&s=62277739ea5e71fd1e80e03447caa9bb" width="2046" height="542" data-path="images/crash-lifecycle.webp" />

1. **Accident Event:**\
   A crash occurs and is reported to local or state authorities.

2. **Report Filed:**\
   Police or involved parties file detailed reports documenting the crash. We do not collect crashes that are not reported to authorities.

3. **Data Acquisition & Ingestion:**\
   We collect and ingest the reports from various state agencies, municipalities, and other sources.

4. **Standardize & Clean:**\
   Data is normalized and cleaned to ensure consistency across sources for easier analysis and API integration.

***

## Data Dictionary

We curate two specialized datasets, outlined below, to capture critical insights from crash events and vehicle involvement. These datasets, combined with other sources, power our advanced risk models.

<Tabs>
  <Tab title="Crash Data Overview">
    Provides a broad view of each crash event, offering insight into crash circumstances, location, and outcomes.

    | **Column**                 | **Explanation**                                                |
    | -------------------------- | -------------------------------------------------------------- |
    | Timestamp                  | Exact date and time when the crash occurred.                   |
    | Manner of Collision        | Describes how the vehicles collided (e.g., head-on).           |
    | Location (Lat / Long)      | Geolocation of the crash scene.                                |
    | Intersection Type          | Type of intersection (if applicable) where the crash occurred. |
    | Severity                   | Describes how severe the crash was (e.g., fatal, minor).       |
    | Number of Fatalities       | Number of people killed in the crash.                          |
    | Number of Serious Injuries | Serious injuries requiring emergency medical care.             |
  </Tab>

  <Tab title="Vehicle Level Data">
    ## Vehicle Level Data

    This dataset focuses on individual vehicles involved in accidents, providing granular information relevant to the severity and nature of each crash.

    | **Column**     | **Explanation**                                              |
    | -------------- | ------------------------------------------------------------ |
    | Owner ZIP      | State where the vehicle's owner is registered.               |
    | Impact Area    | Area of the vehicle that sustained the impact (e.g., front). |
    | Damage Scale   | Severity of the vehicle damage (minor, moderate, severe).    |
    | Vehicle Make   | Make of the vehicle.                                         |
    | Vehicle Year   | Manufacturing year of the vehicle.                           |
    | Towed          | Indicates whether the vehicle had to be towed post-crash.    |
    | Commercial Use | Whether the vehicle was used for business purposes.          |
  </Tab>
</Tabs>

***

## Data Accuracy

We rely on the Fatality Analysis Reporting System (FARS), a nationwide database of fatal crashes with strict reporting standards, to ensure data accuracy. Our fatality data consistently aligns with FARS figures, matching within 5% accuracy 88% of the time.

If you come across any inaccuracies in our data, please let us know by [contacting us](mailto:sergey@matrisk.ai).