Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

April 3, 2023

Detecting and Mitigating Traffic


Accidents using Machine Learning
and Traffic Data
Presented by:

Brandon S. Copeland

Omdena Jordan Chapter


Why did we do it?
Why we decided to take on that
topic?
● Traffic accidents are considered the second
leading cause of death in Jordan.
● During the past 20 years, the number of
accidents increased from 15,884 accidents in
1987 to 110,630 accidents in 2007.
● For the same period, the population and the
number of vehicles increased only by
approximately 2 and 3 folds, respectively
Obviously, this situation is not surprising,
because Jordan has not yet applied a
comprehensive strategy to address this
problem.
Detecting and Mitigating
Traffic Accidents using
Machine Learning and
Traffic Data
Data Collection
● We were given three tabular datasets containing records of accidents from different
areas of Jordan: “2021‫م‬.xlsx”, ”2021 ‫اﺣداﺛﯾﺎت‬.xlsx”, and “NEWcrash data2.xls” by the Ministry
of Transportation of Jordan.
● The records contained information about the accidents such as accident ID, accident
location, timestamp, and various other details. One of the datasets was provided in
English, which will be dubbed New Crash Dataset in the following, whereas the other
two datasets were in Arabic and required translation.
● In the course of the project, additional data about weather and holidays in Amman,
Jordan were scraped from the internet to gain additional insights into the causes of
traffic accidents in Jordan.

Detecting and Mitigating


Traffic Accidents using
Machine Learning and
Traffic Data
What did we do?
How does it work?
Non technical description

METHODOLOGY & TOOLS


● Translation of datasets from Arabic to
English using Google Translate API for
Python.
● Data quality check and data
pre-processing.
● Cluster analysis: Scikit-Learn library
● Web scraping
● Code maintenance and version control
with GIT

Detecting and Mitigating


Traffic Accidents using
Machine Learning and
Traffic Data
Project Demo
Mapping the coordinates of accidents allows for the
identification of high-incidence areas, or "hotspots," and
patterns in the frequency and location of accidents. This
information can then be used to make informed
decisions and allocate resources toward preventing
future accidents. Furthermore, mapping the coordinates
can provide insights into potential contributing factors,
such as road design, traffic flow, and weather conditions.
The density plot below reveals several hotspots with
varying intensity, allowing for a clear understanding of
the distribution of accidents from January 2014 to May
2018.

Detecting and Mitigating


Traffic Accidents using
Machine Learning and
Traffic Data
What we learned?
What can be next?
● Importance of Stakeholder Engagement
● Importance of quality data Even after repeated requests to the
Unfortunately, what prevented us from stakeholders for information, we still failed
reaching the optimal result is the to get clarifications or more relevant data
poor-quality data that contains a lot of from the stakeholders.
ambiguity. The data was filled with errors
and inconsistencies, making it difficult to
accurately analyze the information and
use advanced AI techniques.

Detecting and Mitigating


Traffic Accidents using
Machine Learning and
Traffic Data
Big thank you to our collaborators!
● Joseph Antony ● Alexander Lau ● Sahil Ganesh
● Zein Moualla ● Farah Abdou Kadu
● Ousama El Masri ● Sandy Lauguico ● Zaid Rahamneh
● Shwetha Mallikarjun ● Sophia Ibiteim
Hiregowdar Jackrich

Detecting and Mitigating


Traffic Accidents using
Machine Learning and
Traffic Data

You might also like