Abstract: With the melting of sea ice, Arctic sea routes are increasingly used and developed, requiring information to support the management of risks of ship accidents in Arctic waters. We identified latent association rules between ship and accident attributes using accident records from 2008 to 2017, data mining and temporal and spatial analysis. Results show that the number of accidents has been increasing, with most accidents occurring in the harbor of Мурманск in eastern Barents Sea and in northern Norwegian Sea. The small Russian fishing vessel is the dominant vessel type in Arctic waters. Serious accidents in Arctic waters are influenced by several variables such as gross tonnage, flag, vessel type, accident type and accident location. Large ships are more likely to cause serious accidents. Serious accidents of Russian fishing vessels in Arctic waters were often caused by machinery damage, and rarely caused environmental pollution. These findings can be of use to maritime safety administrations and shipping companies in the prevention of accident and risk management of Arctic shipping.

Key words: Arctic shipping, feature analysis of maritime accidents, data mining, spatio-temporal analysis, association rule