Analyzing New York City Motor Vehicle Collisions

https://ray310.github.io/NYC-Vehicle-Collisions/

In this project, we analyze motor vehicle collisions in New York City (NYC) to:

• Understand the magnitude of the problem they represent
• Determine when and where collisions occur
• Evaluate when collisions are elevated relative to traffic levels
• Make it easier to see where serious collisions (collisions with deaths and injuries) are occurring
• Suggest causes and interventions to investigate
• Highlight problematic locations and areas that may not be obvious
  
  

To View Project Website

https://ray310.github.io/NYC-Vehicle-Collisions/

Supplemental Analysis

Vehicle Activity Index

• The vehicle activity index is constructed to identify when collisions are elevated relative to the level of driving.
• The index is based on traffic volumes in the Metropolitan Transit Authority (MTA) Bridge and Tunnel dataset.
• Yearly, monthly, weekly, and daily traffic patterns in the MTA dataset were profiled
• Relationships between toll crossings was evaluated using cross-correlation and by examining the weekly traffic pattern of select crossings. patterns was performed
• Gaps were identified in the MTA dataset and imputed using XGBoost regression.
• Please see the following notebooks:
  • 1b-profiling-mta-traffic-index.ipynb
  • 2a-preparing-mta-data-xgboost.ipynb

Cycling Activity Index

• The cycling activity index is constructed to identify when collisions with cyclists are elevated relative to the level of cycling.
• The index is based on cycling volumes at several bicycle counters throughout NYC in the NYC Bicycle Counts dataset
• The 30+ bicycle counters were reviewed for completeness of data
• The index was constructed using a composite of counters to provide a consistent index going back to 2015
• Yearly, monthly, weekly, and daily traffic patterns in the cycling index were profiled
• Please see the following notebooks:
  • 1c-profiling-nyc-bike-counter-data.ipynb

To Directly View Notebooks with Full Content

Project notebooks with maps can best be viewed using Jupyter's nbviewer.
View project notebooks with nbviewer

Note that some notebooks may be slow to display or may not display well on mobile devices

Data Sources

NYC Collisions
https://data.cityofnewyork.us/Public-Safety/Motor-Vehicle-Collisions-Crashes/h9gi-nx95

NYPD Precinct Shapefiles
https://data.cityofnewyork.us/Public-Safety/Police-Precincts/78dh-3ptz

MTA Bridge and Tunnel Toll Data
https://data.ny.gov/Transportation/Hourly-Traffic-on-Metropolitan-Transportation-Auth/qzve-kjga/about_data

NYC Bicycle Counts
https://data.cityofnewyork.us/Transportation/Bicycle-Counts/uczf-rk3c/about_data

NYC Bicycle Counters
https://data.cityofnewyork.us/Transportation/Bicycle-Counters/smn3-rzf9/about_data

NYC City Council District Shapefiles
https://data.cityofnewyork.us/City-Government/City-Council-Districts/yusd-j4xi

Reproducing Processed Data and Analysis

1. Clone repo
2. Download and save data to local directory, e.g. /data/raw
3. Create and activate project virtual environment
   • Python 3.12.3 is used
   • Virtual environment can be created from either conda_environment.yaml or requirements.txt
4. Update data input and output parameters in process_raw_data.py as appropriate
5. Run process_raw_data.py (this script may take a while)
6. To obtain City Council point of contact information, run src/scrape_city_council.py with required command line arguments
7. Run notebooks