Search. Read. Cite.

Easy to search. Easy to read. Easy to cite with credible sources.

Journal of Applied Sciences

Year: 2013  |  Volume: 13  |  Issue: 16  |  Page No.: 3263 - 3268

An Integrated Data Mining Method and its Apllication

S. Chuanhe, L. Zhongwen and L. Ying

Abstract

In order to address nonlinearity and instability in financial time series, this study is aimed at structuring an integrated data mining method based on Support Vector Machines (SVM) and Wavelet Neural Networks (WNN) and at exploring its application in analyzing financial market interactions. In the proposed methodology, a kind of WNN is employed to select input features for the SVM using variance rating analysis, with the SVM also improved in feature weighting through innovated discounted least square. This model proposed is capable of mining more information contents implied in samples and also enhancing generation ability of the model by taking use of the advantages of WNN and SVM. Empirical results show that the proposed hybrid approach can capture unique interaction mechanisms between financial markets in China more efficiently than other analysis methods.

Cited References Fulltext