Prediction Model For Urban Competitiveness Based On Decision Tree

The decision tree can be constructed very fast and can handle both continuous and discrete data. It also can emphasize more important properties clearly. With these advantages, the decision tree is a dominant method for information retrieval problems.Through the research of urban competitiveness, a city can be aware of its development status and its strengths and weaknesses objectively. The research of urban competitiveness also helps establish the development and competition strategies of the city and promote urban development. Since the reform and opening of our country, the economic development of Shanxi province falls behind the domestic developed regions significantly. One of primary reasons is the decreasing of the urban competitiveness of Shanxi province. Under the strategic circumstance of reinvigorating central regions, it is very important to do research about the urban competitiveness of Shanxi province.Recently, most international and domestic researchers only use traditional statistical methods for research of urban competitiveness. There are few literatures dealing with the urban competitiveness research by using data mining techniques. My thesis studies the applications of the decision tree method from data mining in urban competitiveness research. Furthermore, it investigates urban competitiveness of Shanxi province in both theoretical and practical ways. The investigation combines the qualitative analysis and quantitative analysis methods, and also integrates the dynamic analysis and static analysis together. Then, my results provide some suggestions for improving the urban competitiveness of Shanxi province. The main work of my thesis is listed below:(1) A new theoretical model of urban competitiveness is built by comparing, drawing on and combining different international and domestic urban competitiveness models. The theoretical research also provides the basis for building the evaluation index system for urban competitiveness of Shanxi province.(2) The evaluation index system of urban competitiveness of Shanxi province is established by 6 essential ingredients: economy strength, opening to the outside world, infrastructure, technicians and scientists, resources and environment, government strength; Then the secondary concrete indices are selected based on the 6 essential factors and the reality of Shanxi province.(3) There are 11 cities of Shanxi province that are studied in my thesis based on the new evaluation index system. The different algorithms using the decision tree principles in data mining, e.g. CART, CHAID and C5.0 and so on, are used to classify the 2006 urban competitiveness of those 11 cities. Furthermore, the best decision tree model is selected through these algorithms.(4) The level of urban competitiveness of Shanxi province in 2007 is got by using the data of each index of 2007 in the best decision tree model. The relations between urban competitiveness and other individual competitiveness are also analyzed. These results present the objective comprehensive and accurate evaluation for these main cities of Shanxi province and some suggestions for improving the urban competitiveness of Shanxi province are given.My thesis builds the evaluation index system of urban competitiveness based on the reality of Shanxi province, and then utilizes the decision tree methods of data mining to the classification of the urban competitiveness. After selecting the strongly related properties and reducing the weakly related properties, urban competitiveness in next year is also predicted. Our method combines both qualitative analysis and quantitative analysis, and uses analysis method which treats the expert assessments and scientific computation as complement to each other. Our method is scientifically reliable, simple and practical, thus has lots of practical applications.