Understanding prolonged ferry service interruptions

Repeated cancelled records on one route can form a different pattern from a short weather disruption. But the first or last observed date is not a verified route opening or permanent closure date. Dates absent from the source are not counted as cancellations.

The study’s explicit rule

1. Examine each direction separately, and also each direction paired with an exact departure time.
2. Require at least 14 calendar days, including the endpoints, and at least six distinct observed cancellation dates.
3. Allow no gap longer than seven days between observed service dates.
4. A completed, unresolved or conflicting observation breaks the run. A missing date in the source’s overall calendar also breaks its evidence.

A concrete example: 101 days, 30 observations

An earlier historical review found 30 cancelled observations for the Yenikapı → Bursa 20:30 departure slot between 21 September and 30 December 2018. That interval spans 101 calendar days including its endpoints. It does not mean 101 daily sailings were cancelled, or that the entire route was closed all day. This is why direction and departure slot are treated separately.

Why coincidence with an event is insufficient

A cancellation cluster overlapping a pandemic period, public holiday, weather event or operating change does not establish causation. Dated operator notices and independently checked contemporary sources are still needed. Keeping March–July 2020 outside the initial weather study does not label every cancellation in that period with the same cause.

Describing the past versus predicting a departure

We only learn that a cancellation run lasted two weeks after those observations exist. Using that information in a prediction for its first day would leak future information into the past. The retrospective interruption filter therefore defines a research sample; it is not a live cancellation notice or a certain outcome for a future sailing.

See unfiltered archive statistics · Why final booking counts do not establish causes

Sources and data

İDO: official sailing and ticket information · MGM: Marmara marine forecast

Data methodology and scope