1、时间序列分析实验设计实验报告课程:时间序列分析课程设计 姓名 : 倪瑞 班级:12统计学 2班 学号:12207040217摘要:课程设计的内容主要是通 过对一些典型、 通用的 时间序列模型 的进行创新或者改进,加深学生 对时间序列分析方法中的基本概念和 基本理论的认识,使得学生掌握 时间序列分析的基本方法和技巧。 根 据因材施教的教学方 针,我们并不限制 题目的具体内容, 而是让学生 自己通 过对时间 序列分析 课程的学 习,发现 已有时间序列分析方法的 不足之处后,拟定自己感 兴趣的题目来完成 课程设计的学习任务。用 时间序列预测和分析外来的数据。关键词 :时序图 arma 预测目录;1.
2、1检验时间 序列的平 稳性1.2选择适当模型 拟合曲 线1.3预测未来 值1.4模型 优化 内容如下: 2015 年股市行情分析时间开盘2015-05-22,五4584.98 2015-05-25, 4660.082015-05-26,二4854.852015-05-27,三4932.852015-05-28,四4943.742015-05-29,五4603.472015-06-01, 一4633.12015-06-02,二4844.72015-06-03,三4924.382015-06-04,四4912.952015-06-05,五5016.092015-06-08, 一5045.69201
3、5-06-09,二5145.982015-06-10,三5049.22015-06-11,四5101.442015-06-12,五5143.342015-06-15, 一5174.422015-06-16,二5004.412015-06-17,三4890.55 2015-06-18,四4942.522015-06-19,五4689.932015-06-23,二4471.612015-06-24,三4604.582015-06-25,四4711.762015-06-26,五4399.93请你(1)写出一个恰当的模型 实现上述想法;(2)利用Eiews软件进行分析预测;(3)比较这个模型和其它模型
4、 对数据拟合的差异,给出最优模型,并给出给出最小均方误差预测。0 Senes: OUTPUT Workfik:数摄:Un卅lT, - B XView Proc Object|PropertiesTTPrintName Freeze Default 存Sort Edit-A Smpi-Last updated: 06/28?15-23:33Imported from D:360曲ta唾要數据蛾的文档谶据WsjC5/22/20154584.985/25;20154660,085/26/20154854.855/27/20154932.B55/2 別 20154943745Z29/20156/01/
5、20154603.474633 106/02/20154&44.706/03/20154924.386/04/20154912.956/0520155016.096/08720155045.696/09/20155145.986/10/20155049.206/11/20155101.446/12/20155143.346/15/20155174.426/1620155004.416/17/20154&90.556/18/201541 Ifl.1 时序图如下ViewProcObject PropertiesPrint Name FreezeDefault OptionsSample G图形一直
6、呈增长趋势,无界且不规律,不为平稳序列其自相关图如下View Proc Object Properties Print Name Freeze Sample Genr Sheet Graph StatsDate: Q6/29/15 Time: 00:00Sample: 5/25/2015 6/25/2015Included otiseRations: 25AutocorrelationPartial CorrelationAC PAC Q*Stat ProbII|11 0.654 0 654 12.020 0.0Q1II 1112 0.415 -0 021 17.079 0 000I-113
7、0 327 0 111 20359 0 0001 11匚14 0 179 -0.130 21 384 0.000111匚15 -0.003 -0158 21.384 0.0011 11匚16 *0.133 *0.118 22.009 0.0011 口11 匚17 -0 265 -0.180 24.649 0.001U11 18 -0.348 -0.088 29.444 0.000二11 I9 -0 395 -0112 36 036 0.0001| n110 -0 413 -0 091 43 905 0 000|_11I11 -0.357 0.019 50045 0.0001 1I 1112 -
8、0 294 -0 052 54544 0 000其自相关图不具有短期相关性,进一步说明其不平稳 下面进行adf单位根检验:r3 Series: OUTPUT Workfile: St3s::Untitledn xView Proc ObjectProperties | I Print | NameFreeze |SampleGenr SheetGraphStats IAugmented Dickeyuller Unit Root Test on OUTPUTNull Hypothesis: OUTPUT has a unit root Exogenous: ConstantLag Lengtf
9、i: 4 (Automatic - based on SIC. maxiag=5)t-StatisticProb/Ajgme n憶 d Dickey-Fullertest statist)亡-D 64922108379Test critical values:1% level 5% level 10% level-3808546-30206862.650413MacKinnon (1996) one-sided p-values.Augmented Dickey-FullerTest Equation Dependent Variable: D(OUTPUT) Method: Least Sq
10、uaresDate: 06/29/15 Time; 00:01Sample (adjusted): 5/29/2015 6/25/2015 included observations: 20 after adjustmentsVariableCoefficientStd. Errort-StatisticProbOUTPUTM-0 1695330261132-06492210.5267D(OUTPUT(-1)02047680 325980062816205400有单位根检验P 0.05,接受原假 设,不平稳1,2对该序列进行12步的差分的新序列0 Senes: OUTPUT Workfik:数
11、摄:Un卅lT, - B XView Proc Object|PropertiesTTPrintName Freeze Default 存Sort Edit-A Smpi-Last updated: 06/28?15-23:33Imported from D:360曲ta唾要數据蛾的文档谶据WsjC5/22/20154584.985/25;20154660,085/26/20154854.855/27/20154932.B55/2 別 20154943745Z29/20156/01/20154603.474633 106/02/20154&44.706/03/20154924.386/04/2
12、0154912.956/0520155016.096/08720155045.696/09/20155145.986/10/20155049.206/11/20155101.446/12/20155143.346/15/20155174.426/1620155004.416/17/20154&90.556/18/201541 Ifl.1 画时序图,如下有图可看出一阶差分后其序列始 终围绕0波动,平稳Series: F Workfile :Untitled _ B XVieiv Proc Object Properties Print Name Freeze Sample Genr Sheet
13、Graph Stats IAugmented Dickey-Fuller Unit Root Test on FNull Hypothesis: YF has a unit rootExogenous: ConstantLag Length: 2 (Automatic - based on SIC, maxlag=2)t-StatisticProb.*Augmented Dickey-Fuller test statistic-8.7948440.0000Test criti cal values:1% level5% level 10% level4,297073-3,212696-2747
14、676MacKinnon (1995) one-sided p-vaiues.Warning: Probabilities and critical values calculated for 20 observations and may not be accurate for a sample size of 10Augmented Dickey-FjiierTest EquationDependent Variable: D(YF)Method Least SquaresDate: 06/29/15 Time: Q0:24Sample (adjusted): 5/12/2015 5/25
15、/2015Included observations: 10 after adjustmentsVariable Coefficient Std Error VSiatistic Prob.VariableCoefficientStd. Errort-StatisticProb.YF(-1)-5.7114350649407-8 7948440.0001DYF(-D)2.3954060.4958344.82133600Q29D(YFH)05053250.1848&5273362100340C285.8B612911.2200.0982010.9250R*squared0 997962Mean d
16、ependentar*9021.978Adjusted R-squared0.990943S.D. dependent war163702.6S,E of regression9051213Akaike info criterion21.34836Sum squared resid492E+0BSchwarz criterion21.46939Log likelihood-102741SHannan-Quinn criter21.21559F-statisflc979.3384Durbin-Watson stat1.845358Prot)(F-stati stic)0.000000一阶差分后的
17、自相关图不具有短期自相关性下作 xf二yf-yf(-12)Equation: U1FT TILED Torkf ile: UNTITLED: tIL . _ XViewProcObjectPrintf-Jame iFreezeEstimateForec ast StatsResidsDependent Vanabfe:Method: Least SquaresDate: 06/29/15 Time: 09:50Sample (adjusted): 6/10/2015 6/25/2015Induded observations: 12 after adjustmentsConvergence a
18、chieved after 34 iterationsVariableCoefficientStd EfTor t-StatisticProbC-10917.8816902M -0.0645920.9498AR(1)0.9901010154736 6.3986690.0001R-squared080647SMean depen dent var-27.84917Adjusted R-squared0.787126S.D. dependent var384.2612S E. of regression1772914Akaike in1o citerian13.34448Sum squared r
19、esid314322.3Schwarz criterion13.42530Log liKeiitiood*79 06687Hannan-Quinn enter.13.31456F-statistic41.67383Durbi n-Watsmn stat1.517196Prob(F-statistic)0.000073Inverted AR Roots99预测未来两年的数值= Etiuat ion: UHTITLED Torkf lie: 01TTITLED:U. X. e/ Proc Object Print FJamEeeze Estate ForestReside包逼:W AA匸 AARncKtainplCL S.lfil S 包戈创为015MfUMdi umplt 6.10/2015 6-2&201: 5JOSCODrKti-.E 12g l/ei-i Sks-aTM Ernr 刃7 W1桃紀畑心0住 153SC77AT 心 *沦 EW 3 m&Sipie 皆二卫国 OMn 0021632殉P呻2 a 35i即tr*x napowr 0 03353Cwa-ii-s Pro W C 6li3z-5由其可口呈增长趋势,说明其是单调减序列
copyright@ 2008-2022 冰豆网网站版权所有
经营许可证编号:鄂ICP备2022015515号-1