Application Of Mining Techniques Based On Association Rules In Electric Marketing Analysis

With the deepening market-oriented reform process of electric power industry,power company is now facing the urgent problem of how to establish modern electricmarketing concept, which decides whether it can survival and develop in the marketfull of fierce competition. So far, most methods for electric marketing analysis arefocus on data management and collecting or simple statistics. It's difficult to achievesignificant information for electric marketing decision from increasing marketingdatabase by these theories.Under the background, association rules method is applied to the electricmarketing analysis in this paper. The method is based on the theory of modernMarketing and Data Mining, considering the characteristics of electric industry. Thekey points of the ideas are as follows: the electric market is segmented, according tothe difference of industry and month. the algorithm of association rules is proposed tofind relevancy between electric power consumption and influencing factors, such asprice, precipitation, temperature etc, in the target market. By analyzing the result,effective electric marketing strategy can be drawn up easily.Based on the ideas above, a lot of practical work has been done to apply thetechnique of association rules to the electric marketing analysis. Firstly, datawarehouse with the theme of electric marketing analysis is built. In order to makeanalyze more effectively, the clustering method is used to generalize the original data.Then the association rules algorithm is developed to find frequent items and achievestrong association rules from the data set of electric sale. It's possible that there existsome error conclusions in results, which may misguide the analysis. Therefore, therelativity among the results is checked, which guarantees the veracity of results.According to the practical example given in the paper, it shows that the methodcan acquire association rules fully and quickly. Valuable information can bediscovered from some of results, which is difficult to be found in traditional method.The association rules method is proved helpful in aided decision-making for electricmarketing.