The Analysis And Platform For Meteorological Data Based On Decision Tree Algorithm

There is an inseparable relationship between our life work and weather forecast.Real-time and accurate weather forecast is very important.With the advancement of meteorological-detecting instruments,meteorological-observating techniques and data-saving technology,massive multidimensional and complex meteorological data have been accumulated.On the one hand,it is a great challenge to analyze the logic consist in those data efficiently and accurately.On the other hand,a well operated platform in which meteorological information will be delivered to users in time is indispensable.The meteorological data in this paper based in automatic weather stations in Foshan,Guangdong Province.By studying the C4.5 decision tree algorithm and combination algorithm,the paper structure a rainfull model with which meteorological data can be forecasted and realize the platform releasing information.The details of this paper are as follows:(1)The basic concept,classification,process and performance evaluation of the decision tree algorithm are introduced in this paper.Two improvements of the C4.5 decision tree algorithm are proposed.Firstly,To address the problems that ignoring the influence of different attributes to the classification results,leading to the low accuracy on dealing with multidimensional dataset,the multiple classifiers of C4.5 decision tree algorithm based on the distance weight is proposed.Secondly,for the diverse influence of the minimum number of leaf nodes on the decision tree,the best value for the sample number of leaf nodes is seted.(2)The concept and category of the combination algorithm,especially introduces the implementation process of Bagging and AdaBoost algorithm are introducted.The base classifiers are constructed by the decision tree algorithm and new multiple classfiers will be built by the Bagging and AdaBoost algorithm.The simulated results show that the proposed algorithm has higher accuracy and stability in dealing with multidimensional data sets.(3)The meteorological data from automatic meteorological stations are analyzed.Basing on the data,the paper structure the model with which the rainfull can be predicted.And then the platform can be built.The platform's designing goals,overall framework and modules'function are introduced.This paper users the Vue.js framework based on the MVVM(Controller-Model-View)and the for-end technology such as the JavaScript.