An Analysis of Road Accidental Data Using Clustering and Itemset Mining Algorithms

Abstract

Road accidental detection is one of the emerging issue in recent days, which has been focused by many researchers. Road accident is the major cause for unnatural death, and desirability, which is unpredictable. So, many existing works aimed to develop some prediction approaches for analyzing the real time dataset and predicting the accidental rate for future. But, it limits with the drawbacks like inefficient prediction, reduced accuracy, and increased time consumption. Thus, this paper aims to propose a new prediction model by implementing various data mining techniques. It includes the stages of preprocessing, clustering, and itemset mining. Initially, the dataset obtained from the UCI repository is preprocessed by eliminating the irrelevant attributes and filling the missing values. Then, the density based clustering technique is implemented to group the filtered data into a cluster. After that, the rules are formed based on the support and confidence values for predicting the future. Finally, the frequent items are mined by the use of Apriori algorithm. In experiments, the performance results of the proposed system is validated and evaluated by using various measures such as accuracy, precision, recall, and time consumption