haku: @supervisor Virrantaus, Kirsi / yhteensä: 163
viite: 8 / 163
| Tekijä: | Goite, Habtamnesh |
| Työn nimi: | Visual data mining of winter shipping accident data |
| Julkaisutyyppi: | Diplomityö |
| Julkaisuvuosi: | 2016 |
| Sivut: | 53 Kieli: eng |
| Koulu/Laitos/Osasto: | Insinööritieteiden korkeakoulu |
| Oppiaine: | Geoinformation Technology (IA3002) |
| Valvoja: | Virrantaus, Kirsi-Kanerva |
| Ohjaaja: | Goerlandt, Floris |
| Elektroninen julkaisu: | http://urn.fi/URN:NBN:fi:aalto-201604201867 |
| Sijainti: | P1 Ark Aalto 3674 | Arkisto |
| Avainsanat: | visual data mining winter shipping accident accident analysis Baltic Sea shipping accidents |
| Tiivistelmä (eng): | Maritime transportation involves shipping goods and people across water bodies in vessels. It serves as a crucial social and economic function. Currently more than 80% of the world trade is transported by sea. Among the waterways of the world maritime transportation, Baltic Sea is one of the major maritime highways despite the ice conditions of the area in winters. Winter navigation in the Baltic Sea has become more common in the past few decades in both ship independent navigation and with icebreaker assistance. However, for reasons ranging from harsh ice conditions to human error, a significant number of accidents has occurred in Baltic Sea winter navigation. Consequently, these accidents have caused environmental as well as property damages. The Baltic Sea is prone to critical ecological disasters in case of accidents. As a result, the need for careful risk management is very high. This thesis focused on certain accidents types related to oil spills in the winter periods in Northern Baltic Sea. In this thesis the conditions under which accidents occur were identified by detecting accident locations, operation types of navigational assistance during the occurrence of the accident, environmental conditions and traffic volume in the area. Moreover, correlations between variables in accident occurrences were identified. The data quality and effects it has on results of the data mining was evaluated. The thesis utilized AIS data, accident database and navigational, sea ice and environmental conditions data. Animation video was used to identify accidents and related condition. Visual data mining was used to discover dependencies between the variables. Common visual data mining techniques such as parallel coordinate plot, scatter plot, scatter plot matrix, Cell plots and maps were used to test their suitability in mining of maritime accident data. The quality of the data was evaluated using data quality measures such as accuracy, consistency, completeness and reliability. This thesis revealed that linking AIS data with accident data and other data to understand the navigational condition the vessels had at the time of the accident can only be successful when there is complete and accurate data. Animation video is a useful tool for recreating the situation the accidents occurred in using accident data, AIS data as well as sea ice condition data but the process of watching videos and recording findings is cumbersome and time consuming. Low data volume and poor data quality lead to weak accident analysis results. |
| ED: | 2016-05-01 |
INSSI tietueen numero: 53500
+ lisää koriin
INSSI