摘要
Artificialneuralnetworks(ANNs)havebeenwidelyusedasapromisingalternativeapproachforforecasttaskbecauseoftheirseveraldistinguishingfeatures.Inthispaper,weinvestigatetheeffectofdifferentsamplingintervalsonpredictiveperformanceofANNsinforecastingexchangeratetimeseries.Itisshownthatselectionofanappropriatesamplingintervalwouldpermittheneuralnetworktomodeladequatelythefinancialtimeseries.Tooshortortoolongasamplingintervaldoesnotprovidegoodforecastingaccuracy.Inaddition,wediscusstheeffectofforecastinghorizonsandinputnodesonthepredictionperformanceofneuralnetworks.
出版日期
2003年02月12日(中国Betway体育网页登陆平台首次上网日期,不代表论文的发表时间)