aa组合模型在股票价格预测中应用研究.pdfVIP

27 12 20 10 12 : 1006 - 9348( 20 10) 12 - 036 1- 04 王 晴 (, 2260 10) : , , , , ( SVM ) ( CAR ) ( CA R - SVM ) CAR - SVM SVM , SVM , CA R - SVM , CAR - SVM , , CA R - SVM , , : ; ; ; : T P39 1 : A Research on Stock Price Prediction Based on CombinedM odel WANG Q ing ( N an ong Shipp ing P ro fessiona lT echnology Ins i u e o M anage Inform a ion D epa rm en , J iangsu N an ong 226010, China) ABSTRACT: The paper dea ls w i h s ock price pred ic ion. A mi ing a he prob lem s ha s ock price mi pac fac ors ex is am ong h igh redundancy, non linear charac eris ics, leading o he rad i ionalm e hods, he problem of low accuracy forecas is proposed based on suppo r vec o r m achine ( SVM ) au oreg ressive ( CAR ) s ock mi e pred ic ion m e hod ( CAR - SVM ) . CAR - SVM u ses SVM m ode l o decide he s ock mi e sequence order, hen uses he SVM o decide he in fluence fac ors o f mi e sequence, and ge s s ock mi e sequence o f SVM forecas ing m odel hrough screening cross- valida ion. S ock his ory da a SVM m odel is ve rif ied by he expermi en s. Expe rmi en al resul s show ha , compared w i h radi iona l s ock forecas m e hod, he forecas ing precision o f CAR - SVM m ode l of is high, and he s ock pr ice changes o reflec he nonlinear dynam ic ru le. I has w ide app lica ion prospec s in s ock pr ice predic ion. KEYWORDS: Suppor vec o r m ach ine( SVM ) ; S ock price; Predic ion; Param e er se lee ion , 1 ,