Survey of non-linear trend forecast and application of time series
Xiaodong Qian · Jisuanji gongcheng yu sheji · 2010
Aiming at the problems of time series trend forecast,the techniques of non-linear trend forecast and the primary applications in the financial field are surveyed.The basic principles,the advantages and disadvantages and improvement of main algorithms of neural networks,support vector machines and chaotic theories are primarily introduced.Then these theories which are combined with genetic algorithm or wavelet analysis for forecast are introduced.The non-linear combination of forecast will be a direction of the trend forecast of time series in the future.Finally,the applications of the forecast techniques in the stock price movements and changes,bond price,risk assessment of the insurance companies,and the bank credit risk are introduced.