doi: /j.issn ARIMA ERNNR5 ' T<4= > 01 1), 1), 0 1), 1), 22 1),3 4 1),5 1), 67 2),8 1),9: 1), ; 1)# 1) :!C '&?1 : )

Size: px

Start display at page:

Download "doi: /j.issn ARIMA ERNNR5 ' T<4= > 01 1), 1), 0 1), 1), 22 1),3 4 1),5 1), 67 2),8 1),9: 1), ; 1)# 1) :!C '&?1 : )"

漱蒋
5 years ago
Views:

1 doi: /j.issn ARIMA ERNNR5 ' T<4= > 01 1), 1), 0 1), 1), 22 1),3 4 1),5 1), 67 2),8 1),9: 1), ; 1)# 1) :!C '&?1 : ) ) :!$%9 &' ) #,-./,0, ,,, 456:?1,E mail:lxy_hb007@126.com ARIMA ;BPNN ;ERNN ; [ ;R=; Y;G R G : ARIMA BPNN ARIMA ERNN [ ALR= YG)Y3!UV, G R= YY"< H: _ LR= Y,! SPSS17.0 Eviews8.0 ARIMA,! Matlab8.0 BPNN ARIMA ERNN [, YGUV \ (I:ARIMA ERNN [ [ GY MRE MER MSE RMSE MAE ARIMA BPNN, MRE 5% ( J:ARIMA ERNN [ LR= YY [ GUV" ARIMA BPNN ForecastingincidenceofhepatitisA basedonarima ERNN combina tionmodelinchina XUChunjie 1),YIN Sufeng 1),GUO Chunyue 1),SONGRuirui 1),WANGKeke 1),XIAO Mengying 1),ZHANG & 4\ $1X ; :! 0 SP:201505

2 JournalofZhengzhouUniversity(MedicalSciences) Nov Vol.51 No Xiufeng 1),LIUXiaoyu 2),FANHongmin 1),FENGFumin 1),HUBo 1) 1)DepartmentofEpidemiologyandBiostatistics,SchoolofPublicHealth,NorthChinaUniversityofScience andtechnology,tangshan )RespiratoryMedicine,KailuanHospital,NorthChinaUniversityof ScienceandTechnology,Tangshan Keywords ARIMAmodel;BPNNmodel;ERNNmodel;combinationmodel;hepatitisA;incidence;forecasting Abstract Aim:TocomparetheefectofARIMAmodel,BPNNmodelandARIMA ERNNcombinationmodelinpre dictiononincidenceofhepatitisainchina,andcomparethepredictiveefectamongthem.methods:thedataofinci denceofhepatitisafrom January2004toDecember2015inChinawerecolectedandSPSS17.0andEviews8.0were usedtoconstructarimamodel,matlab8.0wasusedtoestablishbpnnmodelandarima ERNNcombinationmodel.At thesametime,thedatain2015wasusedtoevaluatetheefectofprediction.results:themre,mer,mse,rmse, MAEfitedandforecastedbyARIMA ERNNcombinationmodelwerelowerthanthoseofARIMAmodelandBPNNmodel, andthemrewaslowerthan5%.conclusion:thearima ERNNcombinationmodelforforecastingtheincidenceofhep atitisaissuperiortothesinglearimamodelandbpnnmodel. R= #T! Y>,9 R=Y (, R=Y 9, 3YG; a \]Y.!!G> Y5 ]aa Cb [ (autoregres siveintegratedmovingaveragemodel,arima) 6 > (back propagationnetwork,bpnn) C = [1 6] > Y T S9 Y 6,) SKO 9, < ( RQ Y < #Y IA- K #$,9 A 3! ) T 0 K T O Y ] [7] [G 5 K Y # G # 5 [,Z[! T5 Fc Y -,C 2,"(, T OG / G@ YUV [8] Elman (El manneuralnetwork,ernn) #T,QA Y )/ #N,b X:?a 3M Yc8, a Y A Yc8,! / M # $ [9] 4! L 2004_ 20141R= Y ARIMA ERNN [, [ ARIMA BPNN 20151R= YGY3O, G R= YY"< 1?LM H 1.1?L _ R= ) > G ; ) * > & :?,M )? _ B 1.2 ARIMA ' [1,9 10], Y5 a (Y, ; I- 3Z(AIC) Schwarz 3Z(SBC) (Loglike lihood)! Y " ARIMA M 3, : (ACF) : (PACF)K 3 0a Y,Box LjungQ?173?1O, : 1 3a?1O ARIMA C %A: Ф(B)Ф(B s ) d D S X t = θ(b)θ(b s )ε t ; E(ε t ) =0,var(ε t ) = σ 2,E(ε t ε s ) =0, s te(x t ε t ) =0, s<t C ),B H,ε t,d D a a,p q C b M,P Q a C b M ; d =(1-B) d ; D S =(1-B) SD ;Ф(B)=1- Ф 1 B- Ф p B p ;θ(b)=1-θ 1 B- θ q B q ;Ф(B s )=1- Ф 1 B s - Ф P B Ps ;Θ(B s )=1-Θ 1 B s - Θ Q B Qs 1.3 BPNN ' [2,7] BPNN S 6) Y#T, Q ; ; 6> M, - >6 Q, Q V,Z " M@Y 6> ;, - Y, 6 >,, + Y \,, P Fa, XD - BPNN Y ; :1 K! newf R Y B < 2 K!

3 728!&(& ) " 51 " 6 train 3! sim [< 1.4 ARIMA ERNN R 5 ' [11] ERNN A BPNN Y,, Y Q T Y T 5 X D ERNN Y BaS, Y / E & - Yc8, TE Y Y, R = Y ARIMA, R = Y Y ARIMA [ IAO Y,b ARIMA [. ERNN Y, R= Y Y. ERNN Y Q,, newelm # N # # Q Y " < ERNN 1.5 ' I! F Y G 20151R= Y, \ C (RE), (MRE), Y (MER), 5 (MSE), 5 A (RMSE) " (MAE) [7] 1.6 Z X! SPSS17.0 Eviews8.0 ARIMA,X! Matlab8.0 BPNN ARIMA BPNN [ 2 (I 2.1 ARIMA '(I! Eviews8.0 -AL, -A ADF= ,P=0.3642, LR= F B D=1Ya d=1 Y a a (Y, D@ Y5, [ a Y ACF PACF,, : Y C ", [ GY Z,DTY " ARI MA(1,1,0) (0,1,1) 12, 5; (1-B)(1- B 12 )X t =( B 12 )ε t /( B) Y ACF PACF,,+, 1, a C : AR1=-0.445(t=-4.384,P<0.001),a b : SMA1=0.739(t=4.013,P<0.001), AIC = ,SBC =8.415,Log likelihood = Box Ljung Q =10.753,P = 0.770,! XY G 20151R= Y,G + 1" 3 1 ARIMA(1,1,0) (0,1,1) 12 'A* ACF PACF, B= > (1/10C)3 ' TD ARIMA BPNN ARIMA ERNN [ BPNN '(I R= Y <Y 1,9 BPNN Q N 12 1, N A B C [11] C 5~14 9 5~14. N 2004_ R = Y, C 0.001, 2000 Y>1 tansig, Q logsig,k! trainlm A 2000 H, # N 5~14M, Y

4 JournalofZhengzhouUniversity(MedicalSciences) Nov Vol.51 No (RES) b+# 8M, Y UV X, 12 b T C Y] (Y MSE= ) " BPNN,! G R= Y,G + 1" ARIMA ERNNR5 '(I R = B D=1Ya d=1y a 13N B, _ ARIMA Y [. ERNN Y, 3 YR= Y. ERNN Y Q Q N 1 ; [, N K! A 1~15 Z Y+ K! traingdx learngdm,? 0 BPNN N 1~15M Y RES MSE+ 2 2,# N 11YM2,RES MSE, 98 5 TE C Y] 9 N 11! O X Y ARIMA ERNN [ ARIMA YG,! [ 20151R=Y Y G,G + 1" 7, 2 E F* N RES MSE N RES MSE ' G 3T R= YY [ G U V + 3 3b,ARIMA ERNN [ Y [ G c " ARIMA BPNN, 2 3T R= Y Y [ G, ARIMA ERNN [ Y [ G B, 3 3 H = > H5 T@I C [UV GUV ARIMA BPNN ARIMA ERNN ARIMA BPNN ARIMA ERNN MRE/% MER MSE RMSE MAE MA ERNN [,! L R = Y E G a * [4,12 13] & ' [ G Y MRE 10%MGUV X, 5%MGY 3O I 4 ) ARIMA BPNN [ GY MRE 10%, H 2 3 'H5 T = > 3 J LR=Y Y B ) a a, Ma Y,9 4 Ea ARIMA ERNN "Y ARI a I Y G U V, [ G ARIMA ERNN [,ARIMA ERNN [ [ GY MRE 5%, H ARIMA ERNN [ [ G U V " ARIMA BPNN ARIMA ERNN [ [ G LR= B E XY, ARIMA ERNN [ c X L R = Y Y ( OY BPNN ARIMA ERNN [ ;) YO :1

5 730!&(& ) " 51 " 6 N YO Y,%V N Y;, Kc M Y ; YS, M, <c8 =,9 ], S Z T I Y N 4AB, MSE \ C,O E IY N 2 B< \ J Y,9 # V K# Y 4X! S,ZQ IY# V,!+L savei,gmx!+l load!, b X G V K < 3 ARIMA ERNN [ Y, K! traingdx, K b O YUY, b au Y Q R, trainlm Y P Y,A ERNN ) c K X [11] a 4 [8,14 16] ARIMA ARCHA ARIMA GRNN ARIMA NARNN [ 3! > YYG, 4 V HQMY [ GU V " ARIMA, ARIMA NARNN [ [ G c" ARIMA GRNN [ 9 A# b ARIMA ERNN [ Y [ GUV > N [, ZGLR= Y I [ Z F>,ARIMA ERNN [ A# ; a OYG 3O,9 b ARI MA ERNN [ LR= Y G, LR=Y; :. B, 5; 0/ / Y [9] A :. ) 3! [ M O:1 AN X! Y A,AB B Y, [ Y [G, Z YUcK DKT O, # K OY- 2> S9 Z [.!Y V,A Y A, bcs > Y9 [ 3 2 Y [,!2 G A 3! ;),3K@ Y B Y G [13], c M Y (Y [ 9:; [1] Z,,!,=. Y ARIMA AR= YG)Y3![J].R G ; &,2015, 42(23):4225 [2] GUAN P,HUANG DS,ZHOU BS.Forecastingmodelfor theincidenceofhepatitisa basedonartificialneuralnet work[j].worldjgastroenterol,2004,10(24):3579 [3] WEIJ,HANSENA,LIUQ,etal.Theefectofmeteorologi calvariablesonthetransmissionofhand,footandmouth diseaseinfourmajorcitiesofshanxiprovince,china:a timeseriesdataanalysis( )[J].PLoSNeglTrop Dis,2015,9(3):e [4] GUO X,LIU S,WU L,etal.Applicationofanovelgrey self memorycouplingmodeltoforecasttheincidencerates oftwonotifiablediseasesinchina:dysenteryandgonor rhea[j].plosone,2014,9(12):e [5],\ Z,,=.ARIMA A AIDS 8 G)Y3![J].!&(& ),2015, 50(2):160 [6] &!,M ",? +. [ ^ = ( Y G 4[J]. & D,2014,39(1):52 [7] ZHANG X,LIU Y,YANG M,etal.Comparativestudyof fourtimeseriesmethodsinforecastingtyphoidfeverinci denceinchina[j].plosone,2013,8(5):e63116 [8] YANW,XU Y,YANG X,etal.A hybridmodelforshort term bacilarydysenterypredictioninyichangcity,china [J].JpnJInfectDis,2010,63(4):264 [9] LIUL,LUANRS,YINF,etal.Predictingtheincidenceof hand,footandmouthdiseaseinsichuanprovince,china usingthearima model[j].epidemiolinfect,2016,144 (1):144 [10],,. ' ARIMA ARIMA GRNN [ A# Y)YGUV [J].) D,2009,30(9):964 [11]$%. MATLABR2007 R [M]. : : Q,2007:111 [12]ZHOULL,YULJ,WANGY,etal.Ahybridmodelforpre dictingthe prevalence ofschistosomiasisin humansof QianjiangCity,China[J].PLoSOne,2014,9(8):e [13]Z,!,E6,=.ARIMA C A YG)Y3![J].) > D,2016,20(3):303 [14]WUW,GUOJ,ANS,etal.Comparisonoftwohybridmod elsforforecastingtheincidenceofhemorhagicfeverwith renalsyndromeinjiangsuprovince,china[j].plosone, 2015,10(8):e [15]ZHANGG,HUANGS,DUANQ,etal.Applicationofahy bridmodelforpredictingtheincidenceoftuberculosisin Hubei,China[J].PLoSOne,2013,8(11):e80969 [16]ZHENGYL,ZHANGLP,ZHANGXL,etal.Forecastmod elanalysisforthemorbidityoftuberculosisinxinjiang, China[J].PLoSOne,2015,10(3):e ( ) <)

注会：2015考试全攻略

注会：2015考试全攻略司考 :2015 考试全攻略指南针小编整理了有关司考备考的六大问题, 并在此为各位解答, 供大家参考, 希望能够帮助到大家一何时开始准备 2015 年司法考试二报考科目组合三关于学习资源的选择四学习方法五如何执行学习计划六备考