The Analysis And Application With Great Traffic Data Based On Hadoop

In recent years,with the continuous growth of automobile volume,traffic violations and illegal behaviors emerge more frequently,such as accompanying vehicle,taxi strike and drunk driving,which bring huge hidden peril to city security.So learning how to quickly identify those behavior patterns becomes important to the police in early warning and investigation.Besides,due to the ever-growing amounts of traffic data,the traditional methods of dealing with storage space and computational efficiency can't meet the demands of users any more.Therefore,this paper focuses on two traffic behavior patterns: “car goes with car frequently”——regarded as “accompanying”,and “taxi strike”——regarded as “probe aggregation” in this paper.And we put forward an efficient method of mining accompanying vehicle real-time algorithm in the spatio-temporal big data and the monitoring probe aggregation based on HBase in the traffic data.We also design a platform to analysis and deal with the massive traffic data based on Hadoop.In this paper,the following aspects are discussed as useful exploration:(1)Through the analysis of massive bayonet data and research on pattern of accompanying vehicle,we firstly provide a new definition of frequent itemsets in the mining of accompanying vehicle.Furthermore,we propose the algorithm FSST in the spatio-temporal big data based on MapReduce above,and the experimental results show that FSST algorithm is superior to the traditional Apriori and Sequence-Growth algorithm in accuracy,execution time,memory usage.Finally,the paper puts forward the method to calculate the accompanying vehicle's suspicion.(2)According to characteristics of the probe vehicle aggregation behavior patterns,we define the pattern of probe vehicle aggregation.And referring to grid clustering algorithm and data model of HBase,we present the global all-weather real-time monitoring of probe vehicle aggregation algorithm based on HBase.Finally the experiments based on real,massive traffic data prove the effectiveness of this algorithm(3)We design the traffic-data analysis platform based on Hadoop and achieve the research content above,mainly including big data platform,background applications and foreground display.Owing to the intelligent performance of platform in computing,storage capacity and scalability,the follow-up analysis can also be integrated on the same platform.