證券分析中數(shù)據(jù)挖掘模型的研究及應(yīng)用.pdf

1、廈門大學(xué)碩士學(xué)位論文證券分析中數(shù)據(jù)挖掘模型的研究及應(yīng)用姓名：林香申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：姜青山20070501ResearchandApplicationofDataMiningModelsinAnalyzingSecuritiesNowadays，Anal曲gSecuritiesisthebasicallyresearchingobjectinfinancialanalyzingfieldInthelastfe

2、wdecades，AnalyzingSecuritiesgrowsupquicklyinChinamoreandmorepeopleputtheirmoneyintoSecuritiesmarketwhichisflourishinginrecentyearsAnalyzingSecuritiesSystemislookedforwardtohighq砌itywiththerapiddevelopingofSecuritiesmarke

3、tSesearchingonAnalyzingSecuritiesSystemhasbecomeallimportantissueinfinancialanalyzingareaInAnalyzingSecurities，stockpredictionisallimportantresearching飄Besidesbeingnonlinearnonstationaryanddynamic，financialtimeseriesalso

4、hasspecialproperties，likebeinghighnoisynonnormal，sharppeakedandheavytailedSOstockseriespredictionismorechallengingandhasgreatvaluesin刪calapplicationandbrightprospectinmarketingMeanwhile，similarresearchinstocktimeseriesis

5、alsoanimportantfieldinAnalyziIlgSecuritiesWiththeSecuritiesmarket’SboomstockpricefluctuatescomplicatedlyQuicklyqueryingsimilarstocksfromagreatdealofstockhistorydatabecomesimportantinAnalyzingSecuritiesAccordingtoabovetwo

6、keypoints，WepayattentionontheapplicationoftheforecastingmodelbasedonGABPneuralnetworkTraditionalpredictionmodelsonlyjustworkwellonshorttermpredictions，theyjustfittothedatatheyexperiencedbutnottotheotherkindsoftestingdata

7、andtheresultsfluctuateslowlyincontinuouspredictionInthispaperweproposeastableandefficientmodelbasedOnGABPneuralnetworkforlongtermforecastingthisproposedsystempl鋇：tic岱wellintheexperiencesHerewealsobuildasimilarstocktimese

8、riessearchingdatabaseAfterwithdrawingcharacteristicfxomstocktimeseriesclusterthetimeseries’characteristicvalues、jl，im血zzycmeanclustering(FCM)InFCMweproposeanewvalidityindexwithitweimprovetheclustering’SefficiencyFinallyw