Dataset Card for NYC Motor Vehicle Collisions and Weather Dataset

The NYC Motor Vehicle Collisions and Weather Dataset aims to merge the NYC Motor Vehicle Collisions Data, the NYC Daily Weather Data, and the NYC Borough Data into a single, coherent dataset. This integration will incorporate borough information for each zip code in New York City and enable detailed analysis of the impact of weather conditions on the day of each collision. Such an approach not only facilitates comprehensive collision-weather analysis but also enhances the understanding of collision patterns across different boroughs, offering valuable insights for both dimensions of study.

Dataset Summary

The NYC Motor Vehicle Collisions and Weather dataset, sourced from NYC Open Data and Visualcrossing, provides a comprehensive overview of police-reported motor vehicle collisions in boroughs of New York City, including the Bronx, Brooklyn, Manhattan, Queens, and Staten Island from 2013 to 2023. This dataset includes detailed information such as crash time period, crash date, collision ID, borough, zip code, and precise latitude and longitude coordinates. Each entry also specifies the street name, street type, and the number of persons injured or killed. Additionally, the dataset encompasses the contributing factors for each vehicle involved, the types of vehicles in the collisions, as well as the temperature, precipitation, precipitation type, weather descriptions in NYC on the dates when the collisions occurred.

Supported Tasks

Here are some key tasks that can be conducted using this dataset:

Time Series Analysis: Analyze trends over time in motor vehicle collisions, including fluctuations in the number of accidents, injuries, and fatalities annually or seasonally.
Geospatial Analysis: Utilize the latitude and longitude data to map collision locations, identifying hotspots or boroughs with higher frequencies of incidents.
Statistical Correlation and Causation Studies: Investigate potential correlations between collision rates and various factors like time of day, weather conditions, traffic patterns(type of street), specific locations (boroughs or zip codes), vehicle types.
Machine Learning Predictive Models: Develop predictive models to forecast the likelihood of collisions in certain areas or under specific conditions, aiding in preventive measures.

Languages

English

Dataset Structure

Data Instances

{
  "crash_date": "2021-12-14",
  "borough": "BROOKLYN",
  "zip_code": "11211",
  "latitude": 40.70918273925781,
  "longitude": -73.95682525634766,
  "collision_id": 4486555,
  "crash_time_period": "15:00-17:59",
  "contributing_factor_vehicles": ["Passing Too Closely", "Unspecified"],
  "vehicle_types": ["Sedan", "Tractor Truck Diesel"],
  "number_of_injuries": 0,
  "number_of_deaths": 0,
  "street_name": "BROOKLYN QUEENS EXPRESSWAY",
  "street_type": "ON STREET",
  "weather_description": "Clear conditions throughout the day.",
  "precipitation": 0.0,
  "precipitation_type": null,
  "temp_max": 11.9,
  "temp_min": 6.8
}

Data Fields

Crash Date[Date]: Occurrence date of collision.
Borough[string]: Borough where collision occurred.
Zip Code[string]: Postal code of incident occurrence.
Latitude[float]: Latitude coordinate for Global Coordinate System.
Longitude[float]: Longitude coordinate for Global Coordinate System.
Collision ID[integer]: Unique record code generated by system. Primary Key for Collision table.
Crash Time Period[string]: Classification of crash times into predefined intervals, such as 0:00-2:59, spanning 8 distinct time periods throughout the day.
Street Name[string]: Street on which the collision occurred.
Street Type[string]: On Street /Cross Street/ Off Street.
Contributing Factors[string]: Factors contributing to the collision.
Vehicle Types[string]: Type of vehicles involved in collision.
Weather Description[string]: The weather conditions when collision occurred.
Number of Injured[integer]: Number of people injured in the specified collision incident.
Number of Death[integer]: Number of cyclists killed in the specified collision incident.
Precipitation[float]: The amount of precipitation that fell or is predicted to fall in millimeters when collision occurred.
Precipitation Type[string]: rain, snow, both, or none.
Maximum Temperature[float]: the maximum temperature in degree Fahrenheit when collision occurred.
Minimum Temperature[float]: the minimum temperature in degree Fahrenheit when collision occurred.

Dataset Creation

Curation Rationale

This dataset is curated to shed light on the impact of borough and weather on road safety. It enables a comprehensive analysis of how weather variations and locations influence the frequency and severity of collisions. In addition, it offers insights for enhancing urban planning and road safety measures and serves as a critical tool for conducting time series analysis, geospatial mapping, and statistical studies to identify trends and hotspots. Furthermore, it lays the groundwork for developing predictive models through machine learning, aiming to forecast collision occurrences under specific weather conditions. Ultimately, this dataset aspires to be a cornerstone for data-driven strategies to mitigate traffic-related incidents, bolstering efforts towards safer urban environments.

Source Data

NYC Motor Vehicle Collisions Data, provided by the New York City Police Department (NYPD), is available on the NYC Open Data platform.
NYC Daily Weather Data, provided by Visualcrossing, is sourced from a variety of reputable historical weather data sources, including the Integrated Surface Database for global sub-hourly and hourly observations, MADIS with its extensive meteorological data like METAR, Integrated Mesonet Data, maritime data, and snow data from SNOTEL, the German Weather Service's (DWD) comprehensive database, the Global Historical Climate Network Daily (GHCN-D) for daily summaries, and sophisticated reanalysis data from ECMWF's ERA5 and NASA's MERRA-2.
NYC Borough Data, provided by the Government of New York City, is available on the NYC Open Data Platform.

Personal and Sensitive Information

Care has been taken to ensure that the dataset does not include direct personal or sensitive information about individuals involved in the collisions. While the dataset provides detailed geographic coordinates of collisions, it does not include names, addresses, or any other information that could be used to identify individuals involved. Users of the dataset are urged to follow ethical guidelines and privacy laws when analyzing or sharing insights derived from this data.

Considerations for Using the Data

Social Impact of Dataset

The NYC Motor Vehicle Collisions and Weather Dataset, a fusion of NYPD's collision data, NYC government’s borough data, and Visualcrossing's weather insights, offers a vital resource for understanding the interplay between weather conditions and road safety. Its comprehensive analysis potential enables urban planners and researchers to devise strategies aimed at reducing traffic incidents, thereby enhancing public safety. This dataset represents a significant step towards a more data-informed approach in urban safety and planning, while maintaining a strong commitment to ethical data use and privacy.

Other Known Limitations

Incomplete Geographical Data: A notable limitation of this dataset is the occasional absence of key geographical details such as zip codes, geocodes, borough names, or specific street types (on-street, cross street, off-street). This missing information can hinder the accuracy of geospatial analyses and may lead to an incomplete understanding of collision distributions and patterns within the city.
Unspecified Contributing Factors: The dataset sometimes lacks specificity in detailing the contributing factors for vehicle collisions. Instances where these factors are labeled as 'unspecified' or are missing can lead to challenges in accurately determining the causes of accidents. This lack of detail may impact studies focused on understanding and mitigating the root causes of collisions.
Generalized Weather Data: The weather data included is based on daily records, which might not precisely reflect the weather conditions at the exact time of each collision. This temporal mismatch can introduce biases in analyses that aim to correlate specific weather conditions with the occurrence of road incidents. As a result, conclusions drawn about the impact of weather on collision rates and severity might be less accurate or comprehensive.

Additional Information

Contributions

This dataset was made possible through the invaluable contributions of NYC Open Data and the New York City Police Department (NYPD), providing extensive collision and borough data, and Visual Crossing, for their comprehensive weather data. I extend my deepest gratitude to these organizations for their pivotal role in enabling this research and for their commitment to open data accessibility.