| The growth of economy can promote the development of cities in various aspects,while the great development of cities can also contribute to the development ofeconomy. Therefore, the study of the economic development of cities will have adecisive significance. With the improvement of quality of life and the greatdevelopment of economy, people start to pay attention to the trends of economy, andmore and more scholars devote themselves into the boom research of economicsituation and its future development. However, the economic system is so complex andchangeable that it is difficult to explore its nature.As the number of variables used in economic forecast is usually large, it is easy toproduce the problem of the multicollinearity. Hence, the idea of extracting thecomponents from the variables emerges to eliminate these effects. Gathering thecorrelation analysis, extracting the components, regression and many other advantages,Partial Least Squares (PLS) has become a central issue nowadays. In recent years,machine learning began active. Extended from the initial classification to the regression,the merging Support Vector Machine (SVM) shows its unique advantages and is beingpursued by many people.This paper introduced the research background and related knowledgeabouteconomic forecasting of the city in the first. Next, the paper began to introduce thetheories of PLS and SVM for the following researching. Then, followed by theempirical analysis, we established the PLS-SVM model based on the relevant data of1997-2012, and we predicted the GDP value in2013. State-of-the-art fitting results areobtained according to the forecast error. Then, this paper built the PCA-SVM modelsimilar to our model and the only SVM model for comparing and analyzing with ourmodel. Finally, the conclusion that the fitting results of the PLS-SVM model is betterthan the other two models is done.This paper has good practicability and applicability, and it is worthy of expansionand promotion. |
PDF Full Text Request