The Integration Of Based On Data Mining Prediction Method For The Forecast Of Precipitation In Northeast China

Weather forecasting is the application of the necessary tools for our life, the weather forecast is also a kind of weather prediction, you can decide whether or not to bring rain gear, considering whether to go out according to the tips the of forecast, etc. Weather forecast information is according to the weather forecast, the weather, the principle and method of dynamic meteorology, statistics, and so on, for a particular area or a particular position to make a quantitative forecast and reasonable speculation. Weather forecasting is an important goal of atmospheric science research. The weather forecast for people's life has important significance.Nowadays, along with the progress of science and technology, weather forecast methods are varied, but for every single forecasting method, the result is not the same. This paper also studied the traditional integrated forecast method, on this basis, we compare the advantages and disadvantages of various integration method, this paper proposes a new integrated forecast method based on data mining, artificial neural network integrated forecasting method. The purpose of this study is superior to the ordinary numerical integration method. This article selects four kinds of single forecasting method, artificial neural network prediction respectively, and the method of multiple regression, the SVM prediction and RBF neural network prediction method. First three chapters in this paper involves the introduction of data mining related knowledge, and four kinds of single forecasting method of simple calculation process and principle. In this paper, the fourth chapter puts forward artificial neural network integrated forecasting method.Through data comparison, the proposed neural network integrated forecasting method is superior to the results of this paper involved each single forecast results. The traditional integrated forecast method of weights is more rely on experience and summary, and integration in this paper can avoid the traditional prediction weights selection in the process of difficulties.This article mainly has carried on the following work:? In this paper, the selection of four single prediction methods are introduced, and experimental comparison.? In this paper studied the traditional forecast methods on the basis of comparative analysis, discover the existing problems and deficiencies in the meteorological forecast, puts forward a new integration for its lack of prediction methods, and through experiment data comparison and verification, which proves that the proposed integration of artificial neural network prediction method has higher reliability.