When I do ARMA(2,2) using one lag of LCPIH data This is eview result> > *Dependent Variable: DLCPIH > **Method: Least Squares > **Date: 08/12/11 Time: 12:44 > **Sample (adjusted): 1970Q2 2010Q2 > **Included observations: 161 after adjustments > **Convergence achieved after 14 iterations > **MA Backcast: 1969Q4 1970Q1 > ** > **Variable Coefficient Std. Error t-Statistic Prob. > ** > **C 0.003361 0.001814 1.853352 0.0657 > **DLCPIH(-1) -0.100150 0.053160 -1.883917 0.0614 > **DLCPIH(-2) 0.870456 0.052466 16.59075 0.0000 > **MA(1) 0.532252 0.100110 5.316678 0.0000 > **MA(2) -0.379383 0.099535 -3.811566 0.0002 > ** > **R-squared 0.512067 Mean dependent var 0.014816 > **Adjusted R-squared 0.499556 S.D. dependent var 0.016274 > **S.E. of regression 0.011513 Akaike info criterion > -6.060182 > **Sum squared resid 0.020676 Schwarz criterion -5.964486 > **Log likelihood 492.8446 Hannan-Quinn criter. -6.021326 > **F-statistic 40.92897 Durbin-Watson stat 2.012062 > **Prob(F-statistic) 0.000000 > ** > **Inverted MA Roots .40 -.94 *This is R result> *> dlcpihTsLen <- length(ausT2Ts[,4]) > **> dlcpihArma22Fit <- arima(ausT2Ts[,4], order=c(2,1,2), > xreg=1:dlcpihTsLen) > **> dlcpiArma22hFit <- arima(ausT2Ts[,4], order=c(2,1,2)) > **> dlcpihArma22Fit > * > *Call: > **arima(x = ausT2Ts[, 4], order = c(2, 1, 2), xreg = 1:dlcpihTsLen) > * > *Coefficients: > ** ar1 ar2 ma1 ma2 1:dlcpihTsLen > ** -0.1083 0.8673 0.5263 -0.3716 0.0146 > **s.e. 0.0493 0.0484 0.0894 0.0852 0.0041 > * > *sigma^2 estimated as 0.0001282: log likelihood = 498.38, aic = -984.76** * * * I wonder why the coefficient values are little bit different between them. * * Another thing I wonder is why the AIC value is so significantly different each other*.* * * Please help me, if anyone who have experience both of eview and R is in R community. Thank you. [[alternative HTML version deleted]]
In your first line, you write "ARMA(2,2)." However, what you fit in R is ARIMA(2,1,2). What you fit in eview, I can't tell. Could that explain the difference? HTH, Daniel Young Gyu Park wrote:> > When I do ARMA(2,2) using one lag of LCPIH data > > > > This is eview result > >> >> *Dependent Variable: DLCPIH >> **Method: Least Squares >> **Date: 08/12/11 Time: 12:44 >> **Sample (adjusted): 1970Q2 2010Q2 >> **Included observations: 161 after adjustments >> **Convergence achieved after 14 iterations >> **MA Backcast: 1969Q4 1970Q1 >> ** >> **Variable Coefficient Std. Error t-Statistic Prob. >> ** >> **C 0.003361 0.001814 1.853352 0.0657 >> **DLCPIH(-1) -0.100150 0.053160 -1.883917 0.0614 >> **DLCPIH(-2) 0.870456 0.052466 16.59075 0.0000 >> **MA(1) 0.532252 0.100110 5.316678 0.0000 >> **MA(2) -0.379383 0.099535 -3.811566 0.0002 >> ** >> **R-squared 0.512067 Mean dependent var 0.014816 >> **Adjusted R-squared 0.499556 S.D. dependent var >> 0.016274 >> **S.E. of regression 0.011513 Akaike info criterion >> -6.060182 >> **Sum squared resid 0.020676 Schwarz criterion -5.964486 >> **Log likelihood 492.8446 Hannan-Quinn criter. -6.021326 >> **F-statistic 40.92897 Durbin-Watson stat 2.012062 >> **Prob(F-statistic) 0.000000 >> ** >> **Inverted MA Roots .40 -.94 * > > > > This is R result > > > >> *> dlcpihTsLen <- length(ausT2Ts[,4]) >> **> dlcpihArma22Fit <- arima(ausT2Ts[,4], order=c(2,1,2), >> xreg=1:dlcpihTsLen) >> **> dlcpiArma22hFit <- arima(ausT2Ts[,4], order=c(2,1,2)) >> **> dlcpihArma22Fit >> * >> *Call: >> **arima(x = ausT2Ts[, 4], order = c(2, 1, 2), xreg = 1:dlcpihTsLen) >> * >> *Coefficients: >> ** ar1 ar2 ma1 ma2 1:dlcpihTsLen >> ** -0.1083 0.8673 0.5263 -0.3716 0.0146 >> **s.e. 0.0493 0.0484 0.0894 0.0852 0.0041 >> * >> *sigma^2 estimated as 0.0001282: log likelihood = 498.38, aic >> -984.76* > > * > * > > * > * > > I wonder why the coefficient values are little bit different between them. > > * > * > > Another thing I wonder is why the AIC value is so significantly different > each other*.* > > * > * > > Please help me, if anyone who have experience both of eview and R is in R > community. > > > Thank you. > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide > http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >-- View this message in context: http://r.789695.n4.nabble.com/ARMA-show-different-result-between-eview-and-R-tp3778217p3778299.html Sent from the R help mailing list archive at Nabble.com.
John C Frain
2011-Aug-30 20:55 UTC
[R] Fwd: ARMA show different result between eview and R
---------- Forwarded message ---------- From: John C Frain <frainj at gmail.com> Date: 30 August 2011 21:52 Subject: Re: [R] ARMA show different result between eview and R To: Young Gyu Park <ygpark2 at gmail.com> If you check your manuals you will find that R uses full maximum likelihood while Eviews uses an alternative non-linear method. ?Many programs use different versions of information criteria but these are monotone transformations (ie they all have an extreme value with the same model. Again you need to check your manuals. Best Regards John On 30 August 2011 08:21, Young Gyu Park <ygpark2 at gmail.com> wrote:> When I do ARMA(2,2) using one lag of LCPIH data > > > > This is eview result > >> >> *Dependent Variable: DLCPIH >> **Method: Least Squares >> **Date: 08/12/11 ? Time: 12:44 >> **Sample (adjusted): 1970Q2 2010Q2 >> **Included observations: 161 after adjustments >> **Convergence achieved after 14 iterations >> **MA Backcast: 1969Q4 1970Q1 >> ** >> **Variable ? ?Coefficient ? ?Std. Error ? ?t-Statistic ? ?Prob. >> ** >> **C ? ?0.003361 ? ?0.001814 ? ?1.853352 ? ?0.0657 >> **DLCPIH(-1) ? ?-0.100150 ? ?0.053160 ? ?-1.883917 ? ?0.0614 >> **DLCPIH(-2) ? ?0.870456 ? ?0.052466 ? ?16.59075 ? ?0.0000 >> **MA(1) ? ?0.532252 ? ?0.100110 ? ?5.316678 ? ?0.0000 >> **MA(2) ? ?-0.379383 ? ?0.099535 ? ?-3.811566 ? ?0.0002 >> ** >> **R-squared ? ?0.512067 ? ? ? ?Mean dependent var ? ? ? ?0.014816 >> **Adjusted R-squared ? ?0.499556 ? ? ? ?S.D. dependent var ? ? ? ?0.016274 >> **S.E. of regression ? ?0.011513 ? ? ? ?Akaike info criterion >> -6.060182 >> **Sum squared resid ? ?0.020676 ? ? ? ?Schwarz criterion ? ? ? ?-5.964486 >> **Log likelihood ? ?492.8446 ? ? ? ?Hannan-Quinn criter. ? ? ? ?-6.021326 >> **F-statistic ? ?40.92897 ? ? ? ?Durbin-Watson stat ? ? ? ?2.012062 >> **Prob(F-statistic) ? ?0.000000 >> ** >> **Inverted MA Roots ? ? ? ? ?.40 ? ? ? ? ? ? -.94 * > > > > This is R result > > > >> *> dlcpihTsLen <- length(ausT2Ts[,4]) >> **> dlcpihArma22Fit <- arima(ausT2Ts[,4], order=c(2,1,2), >> xreg=1:dlcpihTsLen) >> **> dlcpiArma22hFit <- arima(ausT2Ts[,4], order=c(2,1,2)) >> **> dlcpihArma22Fit >> * >> *Call: >> **arima(x = ausT2Ts[, 4], order = c(2, 1, 2), xreg = 1:dlcpihTsLen) >> * >> *Coefficients: >> ** ? ? ? ? ?ar1 ? ? ar2 ? ? ma1 ? ? ?ma2 ?1:dlcpihTsLen >> ** ? ? ?-0.1083 ?0.8673 ?0.5263 ?-0.3716 ? ? ? ? 0.0146 >> **s.e. ? 0.0493 ?0.0484 ?0.0894 ? 0.0852 ? ? ? ? 0.0041 >> * >> *sigma^2 estimated as 0.0001282: ?log likelihood = 498.38, ?aic = -984.76* > > * > * > > * > * > > I wonder why the coefficient values are little bit different between them. > > * > * > > Another thing I wonder is why the AIC value is so significantly different > each other*.* > > * > * > > Please help me, if anyone who have experience both of eview and R is in R > community. > > > Thank you. > > ? ? ? ?[[alternative HTML version deleted]] > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >-- John C Frain Economics Department Trinity College Dublin Dublin 2 Ireland www.tcd.ie/Economics/staff/frainj/home.html mailto:frainj at tcd.ie mailto:frainj at gmail.com -- John C Frain Economics Department Trinity College Dublin Dublin 2 Ireland www.tcd.ie/Economics/staff/frainj/home.html mailto:frainj at tcd.ie mailto:frainj at gmail.com