WHITTLE
SAMPLE RUN ONE
# Denotes comments added after the session. # In S-Plus: > X11() # Enable graphics window. > source("whittle.S") # Read in the program. > z <- scan(file="data") # Read in file called data and assign to vector z. > h1 <- whittle(z,model="fgn") # Do the Whittle method with a FGN model and # assign result to h1. #Intermediate results of the minimization. etatry # is H, B is the value being minimized. etatry= 0.5 B= 1.00022927893032 etatry= 0.500000014901161 B= 1.00022925550517 etatry= 0.999 B= 1.02218824080293 etatry= 0.742705939789286 B= 0.884005310343698 etatry= 0.742705053700002 B= 0.884005156062208 etatry= 0.718504838680966 B= 0.880828221477977 etatry= 0.718505168437669 B= 0.880828250207842 etatry= 0.694266514080424 B= 0.879860064932244 etatry= 0.694266980348998 B= 0.879860060786826 etatry= 0.696510889409528 B= 0.879850585969451 etatry= 0.696510445274812 B= 0.879850585779315 etatry= 0.696407782606995 B= 0.879850563700104 etatry= 0.696408237954346 B= 0.879850563701856 etatry= 0.696407327259644 B= 0.879850563699207 etatry= 0.696407077341415 B= 0.879850563699092 etatry= 0.696416131546077 B= 0.87985056386839 etatry= 0.696398023136753 B= 0.879850563868667 theta= 0.140032566394901 0.696407077341415 H= 0.696407077341415 #Estimated H. 95%-C.I. for H: [ 0.683541807805226 , 0.709272346877604 ] 95%-C.I.: etalow etaup #Approximate 95% confidence intervals [1,] 0.6835418 0.7092723 # for H. > h1[[1]]$parameter #h1 contains the parameter estimate. [1] 0.6964071 #[[1]] is necessary for the case where #several subseries are used, to distin- #guish them.
SAMPLE RUN TWO
> h2 <- whittle(data,out=F) #Now we run the estimator with a FARIMA(1,d,1) model (Default). #The intermediate output is not produced. >h2[[1]]$parameter #Output the estimate. [1] 0.6940678 0.0369692 -0.0334319 #Contains the estimated: H, and phi_1 and theta_1 for the #(1,d,1) model. >q() # Quit.
SAMPLE RUN THREE
# Denotes comments added after the session. # In S-Plus: z1 <- rep(0,1000) # Initialize vector. for ( i in 1:1000){ z1[i] <- 1/10*sum(z[(10*i-9):(10*i)])} # Aggregate the vector z by a factor of 10. > h3 <- whittle(z1,model="fgn") # Do the Whittle method with a FGN model and # the aggregated series z1, assign result to h3. #Intermediate results of the minimization. etatry # is H, B is the value being minimized. etatry= 0.5 B= 1.00049318502577 etatry= 0.500000014901161 B= 1.00049316047897 etatry= 0.999 B= 0.990106893837151 etatry= 0.998999062241329 B= 0.990104299949572 etatry= 0.7495 B= 0.862081752334782 etatry= 0.749500294232016 B= 0.862081788105388 etatry= 0.738029903867731 B= 0.860922000220984 etatry= 0.738030172460806 B= 0.860922021808059 etatry= 0.715655615675116 B= 0.860067724273093 etatry= 0.715656076822042 B= 0.860067721857196 etatry= 0.717024808865672 B= 0.860064289453432 etatry= 0.717024362086441 B= 0.860064289356667 etatry= 0.716970451671377 B= 0.860064283504803 etatry= 0.716971326385792 B= 0.860064283507541 etatry= 0.716969576956961 B= 0.860064283505106 etatry= 0.716970100010298 B= 0.860064283504558 etatry= 0.716979577031209 B= 0.860064283683035 etatry= 0.716960622989387 B= 0.860064283683146 theta= 0.13688348209654 0.716970100010298 H= 0.716970100010298 95%-C.I. for H: [ 0.676137287998876 , 0.75780291202172 ] 95%-C.I.: etalow etaup #Approximate 95% confidence intervals [1,] 0.6761373 0.7578029 # for H. > h3[[1]]$parameter #h3 contains the parameter estimate. [1] 0.7169701 #[[1]] is necessary for the case where #several subseries are used, to distin- #guish them.