Link to MyST Deployment: MyST Deployment

Link to DOI:

TIMS Bicycle Crash Analysis¶

Overview & Motivation¶

This project analyzes TIMS (Traffic Information Management System) data related to fatal traffic crashes. The motivation behind this project is to better understand patterns and trends in fatal crashes, with a particular emphasis on geographical clustering. Identifying spatial patterns can help highlight high-risk locations and support traffic safety planning and policy decision-making.

Analysis Conducted¶

The analysis in this project includes:

• Data cleaning and preprocessing of TIMS fatal crash data

• Exploratory analysis of crash trends and characteristics

• Geographic and spatial clustering to identify high-risk areas

• Visualization of fatal crash locations and spatial patterns

All analysis is performed using reproducible Jupyter notebooks.

Running the Analysis¶

Execute All Notebooks¶

To run the full analysis pipeline from scratch:

make all

This installs the local project package and executes all Jupyter notebooks.

Build HTML Documentation¶

To generate the HTML report using MyST:

make html

The output will be available in the _build/ directory.

Cleaning Generated Files¶

Remove all generated outputs and build artifacts:

make clean

Run Unit Tests¶

Run all the necessary unit tests for utility libaries.

make test

Automation & Help¶

To view all available Makefile commands:

make help

Requirements¶

• Conda

• Bash-compatible shell

• MyST for HTML documentation generation

Data and Methods¶

This project uses traffic collision data from the Transportation Injury Mapping System (TIMS) at UC Berkeley, which is derived from the California Statewide Integrated Traffic Records System (SWITRS).

We analyze spatial crash patterns using Kernel Density Estimation (KDE) and DBSCAN clustering, and build predictive models using Random Forests. The analysis is implemented in Python using pandas, NumPy, GeoPandas, and scikit-learn.

Notes¶

This project is designed for reproducible, automated data analysis using Conda, Make, and Jupyter notebooks.