Research On Mining And Belief Network Visual Representation Of Association Rules Of Airport Noise Sets

The airport noise results from various influence factors synthetically. Mining the relationship between the airport noise and the noise influence factors from the airport noise data sets can help to predict airport noise influenced by different combination of the factors scientifically and make the effective noise reduction measures. Since there exists a nonlinear relation between the airport noise and its influence factors, it is difficult to establish a corresponding function model. In this paper, every noise influence factor is analyzed qualitatively and quantitatively, and association rules between the airport noise and its influence factors are mined, and then the association rules are visually represented in the form of belief network.After analyzing the existing interest measures of association rules, this paper proposes an idea to void mining the useless rules through the introduction of the antecedent and consequent of interesting rules. Based on this idea, the BACIRC-FP-Growth-Cube(Based on Antecedent and Consequent of Interesting Rules Constrained FP-Growth-Cube) algorithm is given, in order to overcome the weakness of the most current algorithms which just use interest measure as a filter. In BACIRC-FP-Growth-Cube algorithm, FP-Growth-Cube algorithm is first used to mine frequent itemsets, and then the valuable association rules are generated directly by matching the antecedent and consequent of interesting rules with frequent itemsets to avoid the blindness of unsupervised learning. Finally, the proposed algorithm is applied to mine association rules between the noise influence factors and the airport noise. Experimental results show that compared with the FP-Growth-Cube algorithm,the proposed algorithm is more effective and can greatly reduce or even avoid the generation of useless rules.In order to express all airport noise influence factors synthetically, this paper proposes a method of belief network visual representation of association rules to overcome the weakness of association rules which can not be used to express the links among different rules. The mined association rules between the airport noise and its influence factors are intuitively represented in the form of belief network after learning the network structure and the conditional probability distribution. The association rules express knowledge of data sets,therefore, the rules can provide effective prior knowledge for building a acceptable belief network Compare with the authoritative noise prediction software INM, the experimental results show the feasibility and effectiveness of the method from the viewpoint on the aircraft type selection, one of the noise reduction measures.