Spatial Data Mining Method Based On Support Vector Machine And Its Application On Economic Geography Of Tourism

This Dissertation is based on the economic characteristic of tourism geography,the needs for economic analysis management of tourism geography. Using the theoryand method of classification or regression algorithms for spatial data mining which isthe application of support vector machine, systematic researching on the spatial datamining method of support vector machine and its application on tourism geographyeconomy, some innovation achievement have been accomplished, listed below:(1) Define severral SVM algorithms. Using combinatorial optimization and leastsquare method, and Multi-class Support Vector Machine method, theMC-COLS-SVM classification algorithm has been raised; using combinatorialoptimization method, reducing limitation and complexity of problems, thecombinatorial optimization COLS-BSVR regression algorithm has been raised.Coming up with the feature selection algorithm of support vector regression, and alsoconducted the confirmed analysis.(2) Applying the SVM classification and regression method and theory ontospatial data mining, the spatial data mining theory and method system has beenconstructed. The working procedure and framework of spatial data mining have beendesigned and built, based on the supporting vector machine. Using MC-COLS-SVM(Multi-class Combinatorial Optimization Least Squares Support Vector Machine)optimization idea, the spatial data classification algorithm has been designed.Referring to the idea of COLS-BSVR (Combinatorial Optimization Least SquaresSupport Vector Regression), the spatial data regression algorithm has been designed.(3) Define the index of events and policies and the index of scenic areadistribution. They are widely used in the tourism geography economy. By means oftime series analysis and statistical analysis, focusing on the index of scenic areadistribution, the influence of GDP, CPI on tourism geography economy, and theimpact of events, policies, we have analyzed and extracted the characters of tourismgeography economy. The corresponding tourism geography economy database hasbeen built.(4) The design of the forecasting model of tourism geography economy. Bymeans of COLS-BSVR support vector regression algorithm, creatively built theanalytical forecasting math model which is based on supporting vector machine forthe tourism geography economy. The data structure mode in data mining has been designed; also the effectiveness of analytical forecasting math model has been proved.(5) The design of the risk management model of tourism geography economy.Referring to the MC-COLS-SVM algorithm, COLS-BSVR algorithm and featureselection algorithm, analyzing and extracting the risk characteristics of tourismgeography economy, creatively built the risk management math model, with itseffectiveness being verified.(6) The design and realization of the forecasting platform of tourism geographyeconomy. This platform has three parts. After collecting and preprocessing mass data,save them into tourism geography economy database, then generate forecastinginformation with the help of supporting vector regression algorithm. This informationwill be used for analysis and decision making.This paper contains51Figures,44tables,116references.