# Port Road Two-Wheeler Commute Risk Simulation

## PURPOSE
A SimPy + NumPy discrete-event simulation of two-wheeler workers commuting daily
along the main road from Kumbhabhishekam Port in Kakinada towards the tidal area.
It models how a given number of potholes and their depths translate into accident
probabilities and delay, reproducing the transcript's core claim: hundreds of
workers on two-wheelers face accidents because of large potholes on the route.

## WHY THIS MATTERS
The video is a short local report, not a data story. Its only quantitative anchor
is qualitative: "hundreds of workers travelling daily on two-wheelers." A
simulation is the honest way to add evidence — you parameterise the crowd, the
hazard density and the encounter probability, then let many random commutes play
out over a simulated day. It teaches that a plausible model beats a vague alarm.

## HOW TO BUILD
Use SimPy for the event loop and NumPy for reproducible randomness.
1. `pip install simpy numpy`
2. Model each commuter as a SimPy process that traverses segments of fixed length.
3. Give each segment a pothole density sampled from NumPy (seeded for repeatability).
4. On each segment, sample an encounter; if encountered, roll for an accident using
   a depth-weighted probability.
5. Run N repetitions and report mean accidents, mean delay and a confidence band.

Core encounter loop:

```python
env = simpy.Environment()
rng = np.random.default_rng(seed)
for _ in range(commuters):
    env.process(commute(env, rng, segments, stats))
env.run()
```

## WHEN TO USE IN PRODUCTION
Use it for road-safety planning: comparing repair scenarios, estimating how many
accidents a resurfacing programme prevents, or justifying budget to a municipal
body. Do NOT use it as a real accident predictor or as a substitute for crash
statistics — it is a planning abstraction. It also should not drive individual
routing decisions.

## HOW IT HELPS STUDENTS
Students learn discrete-event modelling, seeded reproducibility, Monte Carlo
estimation and sensitivity analysis — the toolkit for any "what if" operational
question. They also practise turning a news anecdote into a defensible model with
explicit assumptions.

## STARTER CODE
See `starter/main.py` — a SimPy environment with a `commute` process, a
depth-weighted accident probability, and `# TODO` markers for repair scenarios and
sensitivity sweeps.
