Analysis And Prediction Of Corporation Power Consumption Based On Support Vector Machines

Data Mining is s frontier research topic in the information and database technology. It relates to statistics, AI, fuzzy theory and database technology etc.Clustering analysis is one of the main functions of Data Mining. Support vector machine is a newly developed marginal subject, based on complete statistics learning theory, and has excellent learning performance. It is a new machine learning method. It has become a hot spot in machine learning field.This paper discussed the building model and process of the data warehouse. Based on the data of one of the statistical bureaus in TIANJIN, we discussed data warehouse and data mining technology. On the basis of built data warehouse, we made use of Microsoft assemble model to mine data. Using monthly power consumption of these corporations, we analyzed the relationship between power consumption and value added of corporations, monthly average air temperature, business profit etc. Then we made use of Support vector machine to build short-term forecasting model of power consumption. In the end, we made use of LibSVM to predict and analyze.This paper confirmed one of effective method in short-term forecasting model of power consumption, synthetically applying the method of Support vector machine and mainly considering the influence of value added of corporations, monthly average air temperature, business profit on power consumption. We built model using the monthly datum from 2004 to 2006, and checked the effect of the model using the monthly datum of 2007. The result proved that this method was good at predicting the short-term tendency of power consumption.