Passenger Arrival Modelling with a Poisson Process
An academic project analysing Mumbai Metro Lines 2A and 7 ridership data over 628 days (more than 142 million passenger trips) to understand how passengers arrive.
- Used Poisson process theory and Maximum Likelihood Estimation to estimate arrival rates and expected inter-arrival times across days and months.
- Ran goodness-of-fit tests and dispersion analysis to check the model's assumptions, and found signs of overdispersion in the data.
- Discussed uses in crowd management, train scheduling and infrastructure planning, with non-homogeneous Poisson processes and queueing models as possible next steps.
- Python
- Probability
- MLE
- Goodness-of-fit
Interactive: how a Poisson distribution changes with its rate
An illustrative demo of the maths only. It does not show my Metro dataset.