Font Size: a A A

Forecasting Stock Future Market Using The Least Squares Wavelet Support Vector Machines

Posted on:2009-04-29Degree:MasterType:Thesis
Country:ChinaCandidate:X L DuFull Text:PDF
GTID:2189360275950600Subject:Systems Engineering
Abstract/Summary:PDF Full Text Request
The forecast using Support Vector Machine(SVM) is new data mining technique developed from the 1990s.SVM based on the foundations of Statistical Learning Theory,which is a small-sample statistics and concerns mainly the forecast rule when samples are limited. On account of stock and futures is the complicated nonline system,the conventional forecast skill hardly find out the inherent rule.For the sake of furthermore analyze the rule of stock and futures market,the paper deduce a new means to dope out the market base on SVM and wavelt transform,and analyze the advantage and shortcoming pass experiment. The paper will investigate and discuss in as follows:1.First,research status and the advantage ang shortcoming of method was discussed.And introduce the SVM and wavelt theory and correlation model,which form the textual theoretics basis.2.Building the model of support vector machine using the wavelt kernel,and basing the conditiong of kernel function proved the feasibility of some wavelt kernel function.The paper brought forward the method of construct kernel function,which offer the thereunder for finding out more kernel function.At the same time,analyze the effect and excellent parameter of kernel function.3.We brought forward the forecasting model of Least squares wavelt support vector machines,which is applied in Hushen 300 index and America crude oil futures index.we compare on the forecast effect of standard LSSVM and ANN model,approving the advantage of the model in forecasting.At last,we analysis the model advantage and shortcoming and the further study aspect.
Keywords/Search Tags:support vector machine, wavelet transform, kernel function, embedding dimension, predcition model
PDF Full Text Request
Related items