TY  - CONF
A1  - Mougeot, Mathilde
A1  - Azencott, Robert
T1  - Information theoretical methods dedicated to accidents analysis for GIDAS database
N2  - Nowadays, traffic accidents are recorded in historical databases. Regarding the huge quantity of data, the use of data mining tools is essential to help Experts, for automatically extracting relevant information in order to establish and quantify relations between severity and potential factors of accidents. An innovative approach is here proposed for an in depth investigation of real world accidents data base. Mutual information ratio based on conditional entropies is used to quantity the association strength between an accident outcome descriptor (injury severity) and other potential association factors. Information theoretic methods help to select automatically groups of factors mostly responsible of the severity of accident.
KW  - Analyse
KW  - Datenbank
KW  - Datenerfassung
KW  - Konferenz
KW  - Risikobewertung
KW  - Statistik
KW  - Theorie
KW  - Unfall
KW  - Accident
KW  - Analyse (math)
KW  - Conference
KW  - Data acquisition
KW  - Data bank
KW  - Risk assessment
KW  - Statistics
KW  - Theory
Y1  - 2009
UR  - https://bast.opus.hbz-nrw.de/frontdoor/index/index/docId/480
UR  - https://nbn-resolving.org/urn:nbn:de:hbz:opus-bast-4806
N1  - weitere beteiligte Körperschaften: Université Denis Diderot; University of Houston
ER  -