| Summary: | Data of traffic accident in Southeast Sulawesi has increased every year.
Therefore, traffic accident in Southeast Sulawesi needs to get more effective
handling. Effective handling related to right policies about the management and
traffic engineering. It should be supported by knowledge based on a traffic
accidents database. One of the knowledge that may be got is the characteristic of
severity (dead, seriously injured, lightly injured) of the traffic accident.
This research is to apply Classification based on Predictive Association
Rules (CPAR) algorithm in data base traffic accident, Southeast Sulawesi Police
Department between in the period of 2010 to 2012. CPAR algorithm produces Class
Association Rules (CARs) which is used to describe knowledge about the
characteristics of severity of the traffic accident victims.
The results of experiment shows that the main cause of traffic accident were
human factors (driving under the influence of alcohol and driving exceed the
maximum speed) and environmental physical factors (damage road and elbow
road). Types of accidents (single and head-on) and accidents involving motor cycles
contribute potentially that the victims died. Testing the accuracy using 10-fold cross
validation shows that the average accuracy of CPAR algorithm is 48,75% that is
higher than PRM algorithm 41.13%.
|
|---|