
Certificate of Completion
THIS ACKNOWLEDGES THAT
HAS COMPLETED THE SPRING 2026 DATA SCIENCE BOOT CAMP
Elliot Cheung
Roman Holowinsky, PhD
MARCH 25, 2026
DIRECTOR
DATE

TEAM
Predicting streetcar bunching risk-events from risk-patterns: an ML approach to congestion on the TTC
Elliot Cheung

This project develops a flexible prediction tool for predicting bunching risk-events, from confirmed bunching risk-patterns. In this project a bunching risk-event can be specified by 4 integers (b,s,n,m). Then, a bunching risk-event prediction task is given by:
"given a number of bunching incidents greater than b, observed by a pair of streetcars in the last s stops, predict if the same pair of streetcars will encounter another number of bunching incidents greater than n within the next m stops"
For any risk-event, we call a collection of bunching incidents satisfying the conditional statement of the associated prediction task, a risk-pattern. Then, the prediction task for a type of risk-event is a binary classification problem: for each associated risk-pattern, assign a 1 if the risk-pattern extends to a risk-event, assign a 0 if it does not.
This project collects real-time transit data collected at each streetcar stop for the 506 Carlton Streetcar line on the TTC, from 2024 - 2026
