The use of 'big data' in constructing loss-based performance indicators in the maritime industry

Risk based regulation hinges on good measures of risk. This paper discusses how we combine recorded data of maritime accidents in Norway with exposure data extracted from the Coastal Authority's AIS (Automatic Identification System) databases. Previously few good measures of activity (the number of active ships, nautical miles sailed, port calls etc) have been available for these purposes. Thus, while the accident database maintained by the Norwegian Maritime Directorate (NMD) is quite good, it has been hard to use it to calculate accident frequencies. This paper documents and discusses examples of different ways to normalize accident data.