Road Traffic Collisions
Newcastle Upon Tyne, UK | Jul 2024 - Aug 2024
Principal Investigator Department
Dr. Lee Fawcett Statistics
Research Focus
The research project focuses on reducing road traffic accidents and the resulting casualties, a serious issue that causes significant harm to society and the economy. The project aims to understand how different factors, like location and season, contribute to traffic accidents and how safety measures can change these patterns. By analyzing data from various regions, including the Florida Panhandle, the project seeks to identify the most effective strategies for preventing accidents and improving road safety. The ultimate goal is to save lives and ensure that resources spent on road safety are used in the most effective way possible. This work is critical as it not only addresses public safety concerns but also helps optimize the investment in road safety measures.
Responsibilities
In this research project, I played a key role in analyzing road traffic collision data from the Florida Panhandle. The project's overarching goal, guided by Dr. Lee Fawcett, was to reduce road casualties by understanding how various factors contribute to traffic accidents and how safety measures can influence accident trends. My primary task involved shifting from a Poisson model to an Exponential model to better account for zone and seasonal effects in the data. This switch was necessary because the data provided was based on average collision rates rather than discrete counts, making the Exponential model a more appropriate choice.
I worked with data covering 52 geographical zones in the Florida Panhandle, with accident data spanning from 1960 to 2015. My focus was on fitting a random effects model within a Bayesian framework, using an Exponential distribution to account for the time between collisions. I modified existing R code to fit this new model, replacing the log-likelihood function from a Poisson distribution to an Exponential one. This process required a deep dive into advanced statistical concepts, including Bayesian statistics, the Monte Carlo method, and random effects models.
In addition to model fitting, I conducted exploratory data analysis to identify patterns in traffic accidents across different zones and seasons. This analysis revealed key insights, such as the unusually high accident rates in certain zones and the impact of seasonal variations, which are crucial for targeting road safety interventions.
Through this project, I enhanced my proficiency in statistical modeling, particularly within the Bayesian framework, and gained valuable experience in applying these methods to real-world problems. I also learned the importance of addressing missing data and other data inconsistencies to ensure accurate and reliable analysis. This work has significantly improved my ability to contribute to data-driven solutions for public safety challenges, especially in the context of road traffic accidents.
