TTC-Line1-Optimization
TTC Line 1 — Delay Reduction & Scheduling Optimization
Risk-aware scheduling framework for Toronto’s Line 1 (Yonge–University) subway that combines delay forecasting with MILP-based train frequency optimization.
Topic:
toronto-transit— Toronto Transit Commission (TTC) analytics series.
| Part of a Toronto transit analytics series: ttc-predict | TTC-Line1-Optimization | toronto-transit-sentiment-nlp | interactive-transit-dashboard |
Problem
Line 1 is the TTC’s busiest corridor, but passenger-related delays (overcrowding, alarm misuse, track entry) reduce reliability and increase operational cost.
Approach
Analyzed historical delay data from City of Toronto Open Data, forecast hourly disruption risk with Facebook Prophet, and optimized train frequency using mixed-integer linear programming (PuLP). Compared three scheduling strategies, selecting Modified Option 2 as the best cost–reliability balance.
Results
| Strategy | Avg Delay (min) | Daily Savings |
|---|---|---|
| Option 1 – Min Freq | 4.20 | $47.7k |
| Modified Option 2 | 3.14 | $6.6k |
| Option 3 – Max-Min | 3.00 | $0 |
Modified Option 2 reduced average peak-hour delay to 3.14 minutes while saving approximately $6,600 per day versus baseline scheduling.
Tech stack
Python, Facebook Prophet, PuLP, pandas, Jupyter notebooks
How to run
Recommended (Google Colab): 
- Download the TTC subway delay dataset and upload to
/data/. - Run all cells in the notebook.
Local: clone the repo, install requirements.txt, place the CSV in data/, then run src/forecasting.py and src/optimization.py.
Screenshot / demo
See report/final_report.pdf and notebook outputs for forecast charts and optimized schedules. Project banner: TTC optimization .png.