The Rerearch Of Data Mining Algorithm In Logistics Transportation Based On Association Ruies

Data mining is process that is intend to discovery patterns in large databases. By learning the shortcomings of the existing association rule learning methods, this paper improved the algorithm for generating association rules. Then based the improved algorithm and the data of transportation, a model for transportation analysis and optimization is built in this paper. The main contents and research results of this paper are as follows:(1) Improved the algorithm of generating association rules. In most cases, the algorithms will generate a lot of possible rules, it is difficult to select interesting rules and find the useful patterns form them. In order to solve this problem of association rules mining, this paper puts forward a method that combines two existing algorithms. Firstly, establishing an association network which is composed of association rules. Secondly, a community partition algorithm based on the importance of network nodes is proposed, this algorithm is aim to partition the association network.Finally, we get the classification of the possible association rules. This process may reduce the number of possible rules, and help the researchers to eliminate redundant rules and found interesting patterns or rules in large databases more efficiently.(2) Put forward an improved A* algorithm by association rule learning in this paper. When facing multi factors, the traditional A* algorithm has to compute all the factors. Meanwhile, the weights of all factors are difficult to determine. The improved A* algorithm is proposed by considering association rules. The algorithm selects the main factors from all factors through the support, which is get by association learning. And then set the weights by the correlation strength between the factors.(3) Based on modern logistics data and the algorithm mentioned above, a model for transportation analysis and optimization is created and implemented. Initially, the key operation of GPS data preprocessing is the grid method, which is intended to wipe off the redundant data and offer the road network. Furthermore, a practical and scalable model for transportation analysis and optimization is developed. The model involves multi factors such as time,vehicle types, freight types and so on by extracting the transportation topology network from the historical track transportation data. This model can provide reference and decision support for enterprise transportation planning and transportation regulation of relevant government departments.