Research On Apriori Algorithm In Urban Road Traffic Information Mining

In information age, methods that used to recognize the world such as perceiving,measuring, recording, transmission and storage have got leaping development. Thoughthe size of data present rapid growth pattern, get amount of information order to changethe world is relatively poor. With the development of Intelligent Transportation Systemtechnology, we can get counts of traffic running data by floating car, which equippedwith GPS and communication equipment to collect traffic information. The threeelements of human beings, vehicles and road restrict with each other, build up the urbanroad traffic information. The systems has a characteristic of time-varying, complexityand nonlinear, external and internal factors made the system a high degree ofuncertainty. How to let the traffic original data “make a sound” and get information thatis hidden and valuable is really matter, by sensing, spreading, understanding andresponse.This thesis studies the basic issues of association rule mining, summarizes theclassic association rule mining algorithm. Highlight two kinds of most classic miningalgorithm at the field of association rule mining. Combined with the feature of the urbantraffic information, in view of the poor performance in Apriori algorithm, we makesome research to improve it.By changing the mapping way of database, we avoid the overhead of repeatedlyscanning database in original algorithm. When get the support of each exist candidateitem to determine whether it belongs to frequent items. Since the element in existingfrequent items generate candidate item mustn’t frequent items, with the aid of algorithmpriori knowledge, the element need not to participate in Join in the follow steps. Itoptimizes the Join step. Moreover, when scanning the database, the Apriori algorithmneed to match the candidate items and transactional database. It cost much overhead.We consider import set intersection operation, which is brief, clear and easy tounderstand. Through those three improved strategy, we present the integratedoptimization algorithm. In this paper, the efficiency of optimization algorithmperformance better compared with original algorithm theoretically, and the betteradaptability in the field of urban road traffic information mining.After analysis and study the mining algorithm, many researchers choose mushroomset as test data. And so far, lack of effective association rules mining study on the road traffic congestion, in view of the urban road traffic flow of GPS data match with strictrequirements of association rule mining algorithm. So in this paper, we try to apply theimproved optimization algorithm to the urban traffic information. With the aid of trafficdata extract from the urban’s floating car monitoring center and through data selectionand pretreatment, we mining data by one area and Inter-area at morning rush hour andevening rush hour. Finally, we get the frequent items which meet the constraintconditions. As a result, all of the strong association rules that both exist and significantwere found.Considering the influence of the parameters on the results, we make some furtheranalyze, it also verify that the efficiency of improved optimization algorithm. Analysisthe association rules we got, the result could provide path decisions for the road trafficparticipants, guide them to go out more scientific and achieve traffic forecast andcontrol accurately.In the last part, the research of this thesis is summarized, and the future work isalso discussed.