DefinePK

Abstract

According to the World Health Organization (WHO), road accidents are regarded as one of the leading causes of death. The trend of a road accident can change in future as it is hard to predict the rate at which road accidents are taking place. The road accident leads to an unacceptable loss in terms of property, health and other economic factors. There are instances where road accidents occurred more frequently at a specific location. Some of the road accident features influence road accident to occur frequently. So, it is essential to identify the correlation in various attributes of road accident for predicting road accident. Data mining techniques are widely used to find the correlation in various attributes of the large database. A data mining approach was proposed to characterize road accident locations. In this approach, the Apriori algorithm was applied to characterize locations by generating rules. The Apriori algorithm has high space and time complexity problem and it is also costlier process owing to a large database. In this paper, Frequent Pattern-growth (FP-growth) is introduced for road accident prediction. In FP-growth, the larger feature space is condensed into smaller sub-spaces so that the costly repeated scans are avoided. Then the attributes with high confident values are trained by a decision tree classifier called as J48. It trains and classifies the data as critical and non-critical accident type. Hence by using FP-growth space and time complexity of association rule mining based road accident prediction is reduced and its accuracy is improved by using J48 classifier.