The Implementation of K-means Algorithm for Clustering Traffic Accident Rates on the Highway

Abstract

Introduction :  The increase of population in Palu City has result in increased vehicle ownership and increased the risk of traffic problems such as traffic accidents. Sofar, the accident data in the Palu Resort Police Station has not been fully utilized by the interests of related parties. Therefore, the accumulation of data will be processed by data mining techniques. This study aims to cluster the level of accidents in Palu City based on the age of the perpetrators, where the results of the clustering will be used as consideration for the more targeted socialization of traffic accidents. Based on the results of testing with 2 different centroid initialization methods, the results obtained indicatethat centroid initialization using the ranking method has an SSE value of 233.0690397 while centroid initialization using a random method has an SSE value of 356.42304. It proves that centroid initialization using ranking method has better clustering results compared to centroid initialization experiments using random methods.