Predictive and diagnostic layers computed from the archive: what is about to go wrong, what it costs, and what an operator could do about it. Everything on this page is modelled from observed data and labelled with its method — … days of archive so far, so confidence grows daily. BC Ferries · published AIS · observed
Knowing which sailings will inherit a delay means passengers can be warned before they leave home, not at the berth.
The budget line that justifies (or kills) any operational fix.
Fix the biggest bar first — attacking a 7% cause can't move the network.
| Vessel | Class | n | On-time | Avg delay | Recovery/leg | Median turn | Windy ΔOTP |
|---|
Vessels that add delay every leg need slack scheduled around them; vessels that recover can absorb tight turns.
| Terminal | Deps w/ actuals | On-time dep | Med dep delay | Turn med | Turn p90 | Recovery | Queue %/h |
|---|
Terminals that absorb inbound lateness are the network's shock absorbers — the ones that don't are where berth investment buys the most reliability.
| Day | Posts | Sentiment | Negative | On-time | Sellouts | Gust |
|---|
Operational metrics say what happened; riders say what it felt like. When these two disagree, the gap is the story.
Turns "we should add a sailing" from an opinion into a number someone can challenge.