R in Time Series

CNwt CQmi =
=Q u H xm = x@
u s v
QwBU=@a O}yW jQ@ w ? CavY x=oWv=O
QO xS} w EL=@t
100 Q=DWwv CU=Q} w
|vWwOv|wUwt O}aUO}U
s_mousavi@pwut.ac.ir
1391 R}}=B
?r=]t CUQyi
O Q=DioV}B
1 v tR |=y|QU pw= pYi

| =
1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |= Q v tR | U 11
1 . . . . . . . . . . . . . . . . |rY Q}}e O | = Q s} Q
i C= D w vwQ ' v tR | U U D 111
1 . . . . . . . . . . . . . . . . . . . = Q s} Q
y| U U D 1111
3 . . . . . . . . . . . . . . . . . . | = Q Ov
v tR | U J 2111
5 . . . . . . . . . . . . . . . . |rY Q}}e O
i C= D w vwQ 3111
6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . | = = Q x RH v tR | y| U } D 21
6 . . . . . . . . . . . . . . . ............... = O yp t 121
7 . . . . . . . . . . . . . . . . . . . . . . . . . R = xi w x RHQO y r -t } D 221
7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |oDU@t y 31
8 . . . . . . . . . . . . . . . . . . . . . . T = u}o =} h Qa
v } Q=w w v t } D 131
8 . . . . . . . . . . . . . . . . . . . . |oDU@t T = w h Qay w v } Q=w m } D 231
9 . . . . . . . . . . . . . . . . |oDU@t w T = w w ` w h Qa
yO N w v } Q= m D= @= D } D 331
10 . . . . . . . . . . . . . . . . |oDU@t w x@ =L
yO N U t 1331
10 . . . . . . . . . . . . . . . . . . . =o |oDU@t
Q v y 2331
hr=
16 x = } B |iO=YD |=ypOt swO pYi
16 . . . . . . . . . . . . . . . . . . . . . . . . . . ......... O}iU xiwv 12
16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x Ok t t 112
17 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . h Qa } D 212
17 . . . . . . . . . . . . . . . . . . . . . . . . . . R = x}@
QO |R U W 312
17 . . . . . . . . . . . . . . . . . . Q=ov|oDU@ty w swO x@DQt X= N w 412
19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .| iO=YD uORs o = 22
19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x Ok t t 122
19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . h Qa } D 222
20 . . . . . . . . . . . . . . . . . . . . . . . . . QU =kD Qort w B p v= a 322
21 . . . . . . . . . . . . . . . . . . . | =Y = x@ Q w
iO D uORs o swO D t X= N 422
21 . . . . . . . . . . . . . . . . . . . . . . . . . . . p =i Qort D a 522
21 . . . . . . . . . . . . . . . . . . .......... = x}@ |R U W 622
23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . w} Q w
U oQ D= | yp t= O 32
23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . h Qa } D 132
24 . . . . . . . . . . . . . . . . . . . w} Q w | =DU = | =DU
U oQ D= QO } }= v w } }= 232
25 . . . . . . . . . . . . . . . . . . . . AR(1) O x@ Q w p t swO D t X= N 332
26 . . . . . . . . . . . . . . . . . . . . . AR(1) Ov Q =o |oDU@t } i Q v y 432
26 . . . . . . . . . . . . . . . . . . . . . . | R |oDU@t w ` = } H yO N @ D 532
27 . . . . . . . . . . . . . . . . . . .......... = x}@ |R U W 632
27 . . . . . . . . . . . . . . . . . . . . . . . . xD = Q = O i } VR= @ | yp t 732
27 . . . . . . . . . . . . O
x W |R U = x}@W | yxO=O = Q@ xDi=} Q
VR= @ | yp t = O 832
28 . . . . . . . . . . . . ... xDi=} VR= @ Q AR O

p t % |v=yH =
| tO | U Q 932
30 =DU }= |=ypOt swU pYi

30 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . QLD u}o =} l t =
v t | yp t O 13
30 . . . . . . . . . . . . . . . . . . . w h Qa MA(q) Ov Q
X= N w } D % } i 113
32 . . . . . . . . . . . . . . . = x}@
|R U =o |oDU@t R = =F
W w Q v y % | yp t 213
34 . . . . . . . . . . . . . . . . . . . . . . . . . . . . xD = Q MA i } VR= @ =
| yp tO 23
34 . . . . . . . . . . . . . . . . O = x}@ Q x xD = Q O
x W |R U W | U @ i } VR= @ p t 123
35 . . . . . . . . . . . . . . . . . . . . . . . . . ARMA Ov Q |@} Q } i % =
m D | yp t O 33
35 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . h Qa } D 133
36 . . . . . . . . . . . . . . . . . . . . . . . . O x@ Q w p t swO D t X= N 233
37 . . . . . . . . . . . . . . . . . . . . . . . . . | QH p}rL ARMA @ D D % =
| yp tO 43
37 . . . . . . . . . . . . . . . . . . . . . . . . . Q = x}@ VR= @ w |R U W 143
?
37 . . . . . . . . . . . . . . . . . . . . . . . . . . . p O@ } Q
D Mv | U Q 243
39 =DU}==v |=ypOt sQ=yJ pYi

40 . . . . . . . . . . . . . . . . . . . . . . . . . . . . |rY Q} ARIMA i e p tO 14
40 . . . . . . . . . . . . . . . . . . . . . . . Q O} w Q p =i
j @ r D | U w D 114
41 . . . . . . . . . . . . . . . . . . . . . . . . . . . . |k}ir O D p t 214
42 . . . . . . . . . . . . . . . . . . . . . . . . . . = =F h Qa yp t w } D 314
43 . . . . . . . . . . . . . . . . . . . . . . . . . Q = x}@ VR= @ w |R U W 414
44 . . . . . . . . . . . . . . . . . . . . . . . . . . . . |rY ARIMA = i | yp tO 24
44 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . h Qa } D 124
45 . . . . . . . . . . . . . . . . ............ Q x VR= @ } wQ 224
47 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ARCH = | yp tO 34
47 . . . . . . . . . . . . . . . . . . . . . . . . . SP500 = Q | y| U 134
50 . . . . . . . . . . . . . . . ARCH O h Qa O = = Q O
p t } D %|Q= } B v uO m p t 234
50 . . . . . . . . . . . . . . . . GARCH = O x w Q = \U
| yp t @ \ @ t | y @ 334
51 . . . . . . . . . . . . . . . xD = Q GARCH O
i } VR= @ = x}@
p t w |R U W 434
52 . . . . . . . . . . . . . . . . . . . . . . . SP500 Q Q Q | U @ VR= @ 534
53 . . . . . . . . . . . . . . . . . . . . . . . s}r Q O==
k= | U QO |Q= } B v 634
54 . . . . . . . . . . . . . . . . . = x}@ |v} V} GARCH

|R U W w @ B QO 734
56 `H=Qt
G
Q=DioV}B
OW x=Q= OvDUy x=Q |= OD@= QO w | OD@t xv}tR u}= QO xm |v=Um |}=vW Qw_vt x@ R R Q
Q= i=s v |= @ Q |twta Q=DWwv
" CU= =
|R UxO B =} p@=k u=o}=Q w x xm CiQo Q=Qk Q} R
CQ Y @ TQO x@ R C= } U QO w
http://cran.r-project.org/doc/contrib/Mousavi-R-lang_in_Farsi.pdf
j} Q] R= =Q Q}kL sy O
|=x a w CU= xDiQo Q=Qk u=Ovtxkqa xO=iDU= w
OQ t x WO xDio xDWwv xm OUQ|t Q_v x@
w |t
"O W |Q=RoU=BU =yv R= =H u}ty R= xm OvO=O Q=Qk V} wN C@Lt w h]r OQ tw l}vwQDmr= CUB
EL=@t u=wva Q} R Q =L Q=DWwv u}=Q@=v@ "sO=Di= R OOt x@ X=N EL=@t R= |=xQ=B x}yD Qmi x@ x=Q xt=O= QO xOv@
"OO oQ |t KQ]t xOv@ lOv= Ca=@ OL QO ::: = u}tR
w Q t '| L O Q}O=kt '|v=tR =
| y| U Q Ovv=t R S
QO x } w
x@ EL=@t u}= w O Ww |t s}OkD pYi OvJ QO u K W Q xm OQ=O |v=tR =

| y| U Q x@ X =YDN= u EL@t u}rw=
Q@ Z iQ xm OwW|t u=Wv Q]=N =Hv}= QO "O W w pY=L | D Q Kkvt xDWwv =D Q |t V}=Q} w
OO o w O@=}|t xt=O= Q
Qw t
" CU= xO wtv ?Um =Q R R

Q= i=s v @ Q = x}rw= |}=vW 'xS} w EL=@t ?}kaD Q |t=Qo xOvv=wN xm CU= u}=
|= @
u < =kDQ= w K qY= Q

|= @ |O =yvW}B w Q_v Q=y_= xvwo Qy R= xOvU} wv 'CU}v prN R= |r=N Q =L xR}Hw ,=trUt
"O wtv Oy=wN xO =iDU= w p =@kDU=
|vWwOv|wUwt O}aUO}U
1391 R}}=B u= QyD
O
pw= pYi
|v=tR |=y|QU
|v=tR |QU 1 1
x@ Ov=wD|t O =Wt C@F pta "Ovvm|t Q}}eD u=tR

C= y QO xm CU= |}=yOv} Qi C@F R= OvDQ=@a |v=tR =
| y| U Q
=@ Q |t O}m =D
"OO o =Ut |v=tR pY=wi =@ xDUUo C@F
|w |wQ =Hv}= QP
QO "O } B s =Hv= xDUw}B =} w xDUUo CQ Y w
x@ O =Wt xawtHt
C= y w 1,2, ,n CQ Yw x@ |v=tR xawtHt '?U=vt |v=tR T =}kt w |v=tR O@t
<= ? =NDv=
w |t
"O W xO=O u =Wv x1 x2 xn =} w f xt : t = 1 2 ng pmW
|rYi C=Q}}eD w OvwQ '|v=tR |QU s}UQD 1

1
1
= y|QU s}UQD 1111
u} Qi=Ut O=OaD xm OwW|t xO=iDU= AirPassengers u= wva Q} R R OwN QO |}=yxO=O p}=i R= ::: w s}UQD |=Q@
" CU= xOW C@F 1949-1960 =
xQwO QO Q t u}= "OyO|t u=Wv =Q Qiv Q=Ry ?UL Q@ xv=y=t |rrtr=u}@
%p=Ft
> data(AirPassengers)
1
1391 ' |vWwOv|wUwt 2
> AP <{ AirPassengers

> plot(AP, ylab = "Passengers (1000's)")
O w| x
v= D t w window() = x R =
m OQ=O O Hw s v @ u @ R QO | Qo}O `@=D "CU= q=@ | y )= O
m |= H= Q pY=L 11 pmW
600
500
Passengers (1000s)
400
300
200
100
1950 1952 1954 1956 1958 1960
Time
u} Qi=Ut xv=y=t |v=tR | U Q V}=tv %11 pmW
" O}vm xHwD Q} R p=Ft x@ "OyO CUOx@ =Q |v=tR Q

| U R= |= a x wtHt Q} R l}
> x <{ AirPassengers

> y <{ window(x, 1950, 1951)
" CU= Q} R CQ Yw x@ u xH}Dv xm
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
1950 115 126 141 135 125 149 170 170 158 133 114 140
1951 145
" O}vm xHwD Q} R p=Ft x@ x=t ,qFt 'O}W=@ xDW=O pQDvm p=U R= |rYi |wQ O}y=wN@ Qo =

> y <{ window(x, 1950, c(1951,0))
"CU= Q} R CQ Yw x@ u xH}Dv xm
1950 115 126 141 135 125 149 170 170 158 133 114 140
O}vm xHwD Q} R p=Ft x@ =}
"

> y <{ window(x, c(1950,2), c(1951,3))
3 |v=tR =
| y| U Q 1 pYi
1950 126 141 135 125 149 170 170 158 133 114 140
1951 145 150 178
"OQ= v O M} Q=D xm O}vm xHwD Q} R x = = |v=tR
| v y t | U Q x@ uwvm =
V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11 V12

1 3.59 2.20 4.53 7.01 10.45 43.86 59.97 92.29 47.00 19.16 12.45 8.84
2 9.60 19.12 21.93 24.20 25.22 42.28 37.32 14.11 6.75 4.40 2.62 2.82
3 3.13 3.97 17.52 11.77 9.65 18.26 72.26 32.45 12.09 5.12 3.03 2.46
4 2.76 4.17 6.71 10.57 9.87 7.58 15.45 29.85 4.25 1.74 1.12 1.04
5 1.22 2.81 6.25 9.07 28.12 20.00 36.48 45.32 12.19 2.98 2.85 2.19
... ... ... ... ... ... ... ... ... ... ...
x@ wtv sUQ
"O =Q =yu TBU w xO=O Q= kQ sy QU CWB = =
=Q |Q t | yp U = O}=@ =OD@= xv=y=t =
| yxO=O s}UQD Q
|= @
" CU= Q} R KQW x@ w

pw= CQ Y | y ) = O
m "O=O s =Hv= = u}=
=Q Q m u= Dw |t w
CQ Y wO
> x <{ matrix(scan("F:/R_les/data/ghar.txt"), ncol=12, byrow=T)

> y <{ t(x)
> y <{ matrix(y,ncol=1, byrow=T)
> ts.plot(y)
"CU= Q} R KQW x@ swO CQ Y | y )
m w = O
> x <{ read.table("F:/R_les/data/ghar.txt")

> y <{ as.matrix(y)
> y <{ t(x)
> y <{ matrix(y, ncol=1, byrow=T)
> ts.plot(y)
| = v tR |QU OvJ 2111
O}rwD p}=i 1990 =

p U =D 1958 =
p U R= =}r=QDU= qFt
QO , "O=O u =Wv w w x
s-= D CQ Y @ =Q |v=tR | U Q OvJ u= D w |t
u}= w
"OQ=O O Hw u)D ?UL Q@ C qmW w QD}r uw}r}t ?UL Q@ Q}aWr=<=t 'Ca=U C=wwr}m ?UL Q@ Q
j @
"O W w |t xOv=wN RO )
m \UwD online w x
CQ Y @ | U Q xU
> "http://www.massey.ac.nz/~pscowper/ts/cbe.dat"
> CBE <{ read.table(www, header = T)
CBE1:4, ]
" CU= Q} R w x =
CQ Y @ q @ | y )
m = O |HwQN
choc beer elec

1 1451 96.3 1497
2 2037 84.4 1463
3 2477 91.2 1648
4 2785 81.9 1595
|=Q=O w OvQ=Ov M} Q=D Qo}O CQ a =@ x@ "OvDUy x=t w p U = Ok=i q=@ =

| yxO=O O W w |t x_Lqt xm Qw]u=ty
Q
|= @ "Owtv p}O@D |v=tR | U Q x@ ts() ` = @ D R= xO =iDU= =@ =Q < W= =} u}= O}=@ u}=Q@=v@ "OvDU}v |v=tR Q =D =
| U Q N U
"O}vm xHwD Q} R = O x@ Q=m u}=

| y )
m
> "http://www.massey.ac.nz/~pscowper/ts/cbe.dat"
> CBE <{ read.table(www, header = T)
> Elec.ts <- ts(CBE, 3], start = 1958, freq = 12)
> Beer.ts <- ts(CBE, 2], start = 1958, freq = 12)
> Choc.ts <- ts(CBE, 1], start = 1958, freq = 12)
> plot(cbind(Elec.ts, Beer.ts, Choc.ts))
" CU= q=@ = O
| y ) Q
m |= H= pY=L 21 Q=O wtv
cbind(Elec.ts, Beer.ts, Choc.ts)

10000 14000
Elec.ts
6000
200 2000
Beer.ts
150
100
Choc.ts
6000
2000
1960 1965 1970 1975 1980 1985 1990
Time
|v=tR | UQ OvJ V}=tv %21 pmW
1. trend 2. seasonal variation

5 |v=tR =
| y| U Q 1 pYi
|rY i C=Q}}eD w OvwQ 3111
= =}
,v L= w 3 CQB Q}O=kt '2 |rYi C= Q}}eD '1 OvwQ xmr@ OyO|t CUOx@ Q
=Q | U | wor= =yvD xv |v=tR Q
| U sUQ
|rm Cr=L
' QO w CU= |rYi C= Q}}eD p=U Qy QO wor= Q=QmD q=@ p=Ft QO 'OR U = |t Qy=_ R}v =Q =]N Q}O=kt
u}@D Q
|= @ p t O u} QDxO=U "Ov} wo OvwQ 'OvwW|tv Qy=_ |@ w=vD w x xm |v=tR
CQ Y @ Q
| U QO l}D=tDU}U C=Q}}eD
Q t = `}tHD =@ |rYi C=Q}}eD Q}F-=D u= D w |t 'OvwQ K =w uOQw CUOx@ Q CU= |]N Vy=m w V}=Ri= 'OvwQ
|= @ "
Qy Q}O=kt xYqN "OwW|t s=Hv= aggregate() `@=D R= xO=iDU= =@ R QO Q=m u}= "OQ@ u}@ R= xvq=U Q=t x@ xv=y=t
"OR U = |t u}at =yxO=O R= l} Qy |=Q@ =Q pYi cycle() `@=D w OwW|t XNWt boxplot() `@=D \UwD pYi
> layout(1:2)
> plot(aggregate(AP))
> boxplot(AP ~ cycle(AP))
" CU= q=@ = O
| y )
m |= H= Q pY=L 31 Q=O wtv
OO o Q |t x_Lqt xm Qw]u=ty "O W w |t xOv=wN RO )
m \UwD online w x xv=y=t
CQ Y @ |Q =m}@ =
| yxO=O %p=Ft
aggregate(AP)
5000
2000
1950 1952 1954 1956 1958 1960
Time
400
100
1 2 3 4 5 6 7 8 9 10 11 12
=yv boxplot w x W O `}tHD xvq=U Q

| U V}=tv %31 pmW
ts() ` = @ D R= |v=tR Q
| U x@ =yv p}O@D Q
|= @ w OvDU}v |v=tR | U Q =yxO=O u}= "CU= unemploy Q}eD =
t s v
`}tHD `@=D \UwD xvq=U \UwDt "OO o Q |t x_Lqt u QO freq =v

?w D xQwO w `w W Q QDt=Q=B xm w |t
'O W xO =iDU=
"O Ww |t u}at s}UkD Qorta w
> www <{ "http://www.massey.ac.nz/~pscowper/ts/Maine.dat"

> Maine.month <{ read.table(www, header = TRUE)
3. outliers
> Maine.month.ts <{ ts(unemploy, start = c(1996, 1), freq = 12)

> Maine.annual.ts <{ aggregate(Maine.month.ts)/12
> layout(1:2)
> plot(Maine.month.ts, ylab = "unemployed (%)")
> plot(Maine.annual.ts, ylab = "unemployed (%)")
"CU= q=@ = O
| y )
unemployed (%)
3 4 5 6
1996 1998 2000 2002 2004 2006
Time
unemployed (%)
4.5
3.5
1996 1998 2000 2002 2004
Time
=yv \UwDt w |Q =m}@ Q

| U V}=tv %41 pmW
|v=tR |=y|QU x} RHD 2 1
=ypOt 1
2
1
=ypOt u}=Q@=v@ w |t xOy=Wt |rYi

"O W C=Q}}eD =} w OvwQ =y|QU R= |r}N O
QO ' W x_Lqt ,q@k xm Qw]u=ty
O = |t Q} R
" W @ w x xm 'CU= 4 |atH pOt '|v=tR
CQ Y @ Q
| U xO U = x} RHD l} "OvDUy |}=yxirw-t |=Q=O
xt = mt + st + zt
xirw-t "CU= |rYi C=Q}}eD xOv}=tv t s w OvwQ xv=Wv mt xi w r -t " CU= xOW xOy=Wt | U Q xt =
w u tR t u QO xm
" yOO |t u=Wv =Q =]N R}v t z
u =Wv Q} R CQwYx@ xm 'CU= 5 |@ Q pOt l} ?U=vt pOt 'OwW Qy=_ OvwQ l} CQwYx@ |rYi C=Q}}eD Qo =
"O W w |t xO=O
xt = mt :st:zt 0 0 0
4. additive 5. multiplicative
7 |v=tR =
| y| U Q 1 pYi
Q |t p}O@D |atH pOt x@ |@ Q pOt 'Q}N= x]@=Q

"OO o R= sD} Q=or uDiQo =@ OwW|t x_Lqt xm Qw]u=ty
yt = ln(xt ) = ln(mt ) + ln(st ) + ln(zt ) 0 0 0
= mt + st + zt
R QO =yxirw
-t x} RHD 2
2
1
Qo = "Ovm|t Q
OQw @ l QLDt u}ov=}t VwQ R= xO =iDU= =@ =Q |rYi C= Q}}eD w OvwQ decompose() ` = R @ D ' QO
x@ w Q =F
\ @ t xO=O p t |= @ "OO o Q Q |t O=H}= Q=O wtv xU =@ pmW l} x=ov 'O Q}o Q=Qk plot() ` = p w `=
@ D N=O j i @ D
| =yDv= w |t s}UQD
QO w O W 51 pmW |@ Q w |atH pOt CQ Y @ w x Q} R = O
| y ) =
m QO 'q @ p t j @ =F Q O}rwD
pmW OOQo|t x_Lqt =Hm} |rYi C=Q}}eD w OvwQ Q=Owtv l} |wQ xQNq=@ w OwW|t p=@vO |@ Q pOt =yO)m
"61
Decomposition of multiplicative time series

14000
observed
80002000
12000
trend
1.10 2000 6000
seasonal
1.00
0.90
1.06
random
1.00
0.94
1960 1965 1970 1975 1980 1985 1990
Time
Q
| U | y = xirw-t V}=tv %51 pmW
|oDU@ty 3 1
|oDU@ty u}vJ O}=@ =yv p}rLD w x} RHD Q

|= @ "O @ w Ovy=wN xDU@ty |r=wDt = Q}eDt
| y 'Cq L R== |r}N QO
=
| yxO=O j} Q] R= w O W w |t h} QaD |oDU@ty `@=D \UwD |v=tR =
| y| U QO Q |oDU@ty Q=DN=U Q u}at
"OO o
"OO oQ |t Q O
OQw @ x W x y O =Wt
14000
8000 10000
6000
4000
2000
1960 1965 1970 1975 1980 1985 1990
Time
|rYi C= Q}}eD w OvwQ V}=tv %61 pmW
Tv=} Q=w w u}ov=}t h} QaD 1

3
1
Q}eDt u}ov=}t E (x) u Q =v

}= @ @ " CU= xat=H u}ov=}t `k=w QO 'O W w |t xO=O u =Wv E Q
h L =@ xm | =} Q O}t=
x@ xm CU= w
p L C i= = QLv= `@ Qt u}ov=}t E (x )2 ] ; =@
CQ a "O W w |t xO=O V}=tv R}v = x@ m CU= x
"O Ww |t xO=O V}=tv R}v 2 = @ xm OwW|t xDN=vW x Q}eD t Tv=} Q=w
|oDU@ty w Tv=} Q=wwm h} QaD 2

3
1
"CU= h} QaD p@=k Q} R CQ Y @w x Tv=} Q=wwm s=v x@ |twyit 'O}W=@ xDW=O (x y) Q}eD t wO Qo =
(x y) = E (x x)(y y )] ; ;
R= n O x@
xR= v= |= v x wtv l} Qo = "O Q}o|t O
xR= v= =Q (x y) Q}eD t wO u}@ |]N =@
\ DQ= u= R}t `k=w QO Tv=} Q=wwm
" CU= Q} R w x
CQ Y @ u x]@=Q xm Owtv Q
OQw @ =Q xvwtv Tv=} Q=wwm u= D w |t O}W=@ xDW=O (xi yi)
X
Cov(x y) = (xi ; x)(yi ; y)=n
w -1 u} @ ?} Q u}= "Ovm|t u}at Oa)@ uwO@ w x
CQ Y @ =Q (x y) Q}eD t wO u}@ =@
\ DQ= u= R}t |oDU@ty ?} Q
O
"OQ= v O Hw w (x y) = Q}eDt u}@ |]N \=@DQ= xm CU= |vat u}O@ OW=@ QiY ?} Q u}= Qo =
| y "OQ=O Q= k Q +1
w |t h} QaD Q} R
"O W w x
CQ Y @ (x y) | y= Q}eDt u}@ |va } xat=H |oDU@ty
(x y) = E (x x)(y y )] = (xy)

; ;
x y x y
9 |v=tR =
| y| U Q 1 pYi
"OO o Q |t x@U=Lt Q} R CQ Y @w x xvwtv |oDU@ty ?} Q
Cov(x y)
Cor(x y) =
sd(x) sd(y)
|oDU@tyOwN w Tv=} Q=wmwD= `@=wD h} QaD 3

3
1
w x |v=tR |QU l} u}ov=}t `@=D "OwW|t xDN=OQB u}ov=}t x@ =OD@= |v=tR |=y|QU |oDU@ty h} QaD |=Q@
CQ Y @
O = |t
" W @ t R= |a@=D w CU= Q} R
(t) = E (xt )
Q
OQw @ w CU= 6 =DU}= u}ov=}t QO |v=tR Q
| U x =ov 'OW=@v =
u tR R= |a@=D |va} 'OW=@ C@=F u}ov=}t `@=D Qo =
" CU= Q} R w x
CQ Y @ u |= v x wtv
n
X
x = xt =n
i=1
O = |t Q} R
" W @ CQ Y @ ' w x CU= =DU}= u}ov=}t QO xm |v=tR Q
| U l} Tv=} Q=w `@=D
2 (t) = E (xt )2] ;
w x
CQ Y @ u |= v x wtv OQw Q@ "OOQo|t 2 C = @ F x@ p}O@D 2(t) x =ov 'OW=@ =DU}= R}v Tv=} Q=w QO |v=tR |QU Qo =
"CU= Q} R
P
Var(x) = (xt ; x) 2
n;1
xDU@ty CU= umtt =yQ}eDt =t= "OvOw@ =DU}= Tv=} Q=w w u}ov=}t QO xm O}Owtv x_Lqt =Q |v=tR | U uQ wvm =D
|oDU@ =yv = =
| ys o Cw =iD x@ \ki =yQ}eDt u}@ |oDU@ty xm CU= swO x@DQt | =DU}= |v=tR Q
| U " OvW=@ R}v
" Ov} wo 7 Q}N-=D =Q =yQ}eDt u}@ =
| ys o O= = OaD "OW=@ xDW=O
| U Q Qo = "Ovt=v 9 |B=}B |oDU@ty =} w 8 |oDU@tyOwN =Q hrDNt |=yu=tR QO VO N @ w = Q}eDt l} |oDU@ty
w x
CQ Y @ =Q CU= k R=
|a@=D xm k |va} (acvf) Tv=} Q=wwmwD= `@=D u=wD|t x=ov 'OW=@ swO x@DQt |=DU}= |v=tR
"O wtv u=}@ Q} R
k = E (xt )(xt+k )]; ;
`@=D "CU= E (xt ) = E (xt+k ) = |va } ' CU}v u=tR x@ xDU@=w u}ov=}t =Q} R O
OQ= v u tR = x@ |oDU@ k `@=D
w |t h} QaD Q} R
"O W CQ Y @ w x k Q} = =
N-D @ (acf) |oDU@t w
yO N
k = k2
O = |t
" W @ 0 = 2 Q
= }R ' CU= 0 = 1 x m CiQo xH}Dv u= Dw |t q=@ h} QaD R= xO =iDU= =@
6. stationary 7. lag 8. Autocorrelation 9. Serial correlation
|oDU@t yOwN x@U=Lt 1331
x y O =Wt CiH n 1
; w |t "OW=@ xDUUo |v=tR
u= D Q
| U l} x1 x2 xn O =Wt
x y n O}v m Z i u Q wvm =
|va}
(x1 x2 ) (x2 x3 ) (xn 1 xn) ;
Q
Z i swO Q}eDt u=wva x@ =Q x yO =Wt u}twO w pw= Q}eDt u= wva x@ GwR Qy QO =Q x y O =Wt u}rw= Qo = "CN=U =Q
"CU= Q} R CQ Yw x@ |oDU@ty ?} Q 'O}vm
nP
;1 ; ;
xt x(1) xt+1 x(2)
; ;
r1 = s t=1
nP
;1 ; 2 nP
;1 ; 2
xt x(1); xt+1 x(2);
t=1 t=1
n
P nP
;1
Ovt=v
" O =Wt u}@ |oDU@tyOwN ?} Q
C= y =Q u w CU= x2 = n 1 1 ;
xt x1 = n 1 1 xt
w x ;
u QO m
t=2 t=1
" CU= hrDNt =
| yu tR QO = x Q}eD u} |oDU@t `
t x @ y k=w QO m
w |t
"O W xO =iDU= Q} R pwtQi r1 O = R | = O x n Q
R= W @ nQ @ i m xR= v= @ o=
nP
;1
(xt ; x) (xt+1 ; x)

r1 = t=1
n
P
(xt ; x)2
t=1
n
P
"O wtv x@U=Lt k Q} = =
N-D @ =Q |oDU@tyOwN ?} Q u= D w |t j} Q] u}ty x@ "CU= x = n1 xt u QO xm
t=1
nP
;k
(xt ; x) (xt+k ; x)

rk = t=1
n
P
k = 0 1 2
(xt ; x) 2
t=1
R= =Q |v=tR Q
| U C a = q]= R= |r}N w
CQ Y u}= Q
QO = } R "O m Q x@U=Lt =
=Q y rk O =@ k > n4 Q
} v }O =kt Q
|= @ ,q wtat
"O}yO|t CUO
w |t xDiQo Q_v
"O W QO p
2
n Q= Okt 5% uO @ Q=Ow |vat K]U =@ k r |= xrY=i Q
OQw @ |= @ Q
10 Q=ov|oDU@ty 2331
"O Ww |t xO =iDU= =ypw] QwLt k ?U Q =

L @ yZ a Q Q wLt k r Q=O wtv R= ' |v=tR =
| y| U Q Q}UiD w Q}@aD Q
|= @
Q wLt "CU= Oa)@ uwO@ OQ=O Q=Qk |wQ |oDU@tyOwN Q}O=kt xm =yZQa QwLt "Ovt=v Q=ov|oDU@ty =Q Q=Owtv u}=
10. correlogram
11 |v=tR =
| y| U Q 1 pYi
"OQ=O | Q}oxvwtv |v=tR xrY=i x@ |oDU@ OyO|t u=Wv =Q =yQ}N-=D xm =ypw]

=F
p t u= wvax@ "O Ww |t xO =iDU= R QO lag.plot() ` = @ D R= sy x@ C@Uv hrDNt | y = Q}N-=D s}UQD |= @Q
> lag.plot(LakeHuron, lag=4, labels=F, do.lines=F, diag.col = "red")

Q
|= @ "71pm Q | s} Q Q} = =y = LakeHuron =
W OO o t U DC lag=4 N-D Q J D | yxO=O ' U= w
u J 'q @ p t QO = =F
576 577 578 579 580 581 582
576 577 578 579 580 581 582

LakeHuron
LakeHuron
lag 1 lag 2
576 577 578 579 580 581 582

LakeHuron
LakeHuron
lag 3 lag 4
576 577 578 579 580 581 582
Q}N-=D =
| yQ=O wtv %71 pmW
" CU= Q} R w x
CQ Y @ u |rm pmW w |t
"O W =iDU=
xO acf() ` = R @ D R= QO Q =ov|oDU@ty s}UQD w rk x@ U =Lt
acf(x, lag.max=NULL, type=c("correlation","covariance","partial"), plot=TRUE)

u QO xm
" x
CU= =yxO=O Q
Q=O @ %
| x]@=Q R= xm CU= Z i Q V}B |=Q=O R Q

'OO o h L u t oQ P = w u}= Qo = "CU= Q_v OQ t w Q} = Oa lag.max
N-D O= D %
O = |t
" W @ 10 | =v@t QO u sD} Q=or w y v = x wtv xR= v= O Nx m 'OO o Q | x@ =L 10log(N)
t U t
"correlation" Q P
'OO o h L u t oQ = w u}= Qo = "Ovm|t u}at =Q acf ` = @ D ` v w w CU= = QD =
y xD type
m =Q m R= |= WQ %
" CU= `@=D u}= Z i Q V}B

"OO o Q |t sUQ Q=ov|oDU@ty Q=Owtv CQwY u}= QO w CU= TRUE ZQiV}B |=Q=O |k]vt u=twoQ u}= %plot
"O W w |t A J = =yQ}N-=D |=R= x@ acf Q }O =kt w O W w |tv sUQ xOW xDio Q=O wtv 'OW=@ FALSE = w
u t oQ u}= Qo =
R= =yv x@ |UQDUO Q
|= @ w OO o Q |t xQ}NP acf(x)$acf w x
CQ Y @ |Q=O @ QO Q Q}O=kt u}= xm O}W=@ xDW=O xHwD
k=0 Q o= " CU= acf$acfk+1] CQwYx@ q=@ Q=OQ@ u}=Q@=v@ "OwW|t xO=iDU= Q=OQ@ T}Ov= u=wva x@ Q}N-=D u=tR
" CU= l} Q@=Q@ acf$acf1] Ok Q= t ' W @O =
O}}=tv x_Lqt
" =Q Q} R = O
| y )
m acf ` = =F wv x @ D p t u= a @ u wvm =
> www <{ "http://www.massey.ac.nz/~pscowper/ts/wave.dat"

> wave.dat <{ read.table (www, header=T)
> par(mfrow=c(2,1), mar=c(2,4,2,2))
> plot(ts(waveht), ylab = 'wave height (mm)')
> acf(waveht, main="")
"CU= q=@ = O
| y )
500
wave height (mm)
0
500
0 100 200 300 400

1.0
0.5
ACF
0.0
0.5
0 5 10 15 20 25
u Q =ov|oDU@ty w |v=tR | U Q V}=tv %81 pmW
w |t x@U=Lt Q} R
"O W w x
CQ Y @ k=1 QO acf Q= Okt p@k p=Ft xt=O= QO
> acf(waveht)$acf2]
1] 0.4702564
"O Ww |t x@U=Lt Q} R w x O =
CQ Y @ ' W @ k=1 QO Tv=} Q=wwmwD= x@U=Lt Qw_vt Qo = ,=OOHt
> acf(waveht, type = c("covariance"))$acf2]

1] 33328.39
xrY=i xv=Wv xm OvQ=O OwHw =y k u}}=B r =
w q @ QO u}J]N w x \N
CQ Y @ wO O}vm|t x_Lqt xm Qw]u=ty
O = |t =yxvwtv
" W @ O N
xR= v= u QO xm 'CU= p
2
N
Q@=Q@ %5 uO @ Q=O w |vat K]U =@ w OvDUy u=v}t]=
r
u}@ R= k Q}O=kt Qo = "O Q}o|t Q=Qk xO =iDU= OQwt | U Q |oDU@ty u}}aD |=Q@ u R= xm CU= =D =
|Q N U rk Q }O=kt
x@ O}=@ =Hv}= QO xD@r= Q
"O m OQ =Q %5 K]U QO k = 0 |va } QiY Q
Z i u= D w |t x=ov 'Ot uw Q}@ Q}N= \ w]N
\N wO xrY=i uwQO Q}O=kt xm u swO " CU= |vFDUt xOa=k u}= R= r0 = 1 x m u}= pw= " CW=O xHwD xDmv xU
Q}N-=D 40 |=Q@ ,qFt %5 K]U QO xm u swU "OvQ=Ov |Q=O|vat Cw=iD QiY =@ =t= OvW=@v QiY ,=k}kO CU= umtt
" O}vmv |Qw=O u T U= = Q@ w O}v=Ov =]N pwtWt =Q r0 < RH@ =D
uO i= uw Q}@ =D =
wO ' yxO=O O= OaD =@ ?U=vDt
Q =ov|oDU@ty O
QO " yO |t u =Wv xDUy Vy=m l} w x
CQ Y @ Q =ov|oDU@ty QO =Q O N w =yxO=O OvwQ ,qwtat
13 |v=tR =
| y| U Q 1 pYi
w CU= Q
| U QO x@=Wt Q}O=kt R= |W=v xm C@Ft w nQ @ R Q}O=kt 91 pmW acf(AirPassengers) =
| yxO=O
"O}vm|t x_Lqt =Q "OvDUy syx@ l}ORv = =

| yu tR QO
Series AirPassengers
1.0
0.8
0.6
ACF
0.4
0.2
0.0
0.2
0.0 0.5 1.0 1.5
Lag
AirPassengers =
| yxO=O Q =ov|oDU@ty %91 pmW
=
QO ' yxO=O | U Q xvq=U =v
?w D "O W w |t Qy=_ Q=ov|oDU@ty |wQ ' W @O = xDW=O OwHw R}v |rYi C= Q}}eD Qo =
x]@=Q xm CU= p=Um} Q}N-=D QO |oDU@ty Q=Okt QFm =OL "OwW|t x_Lqt ?w=vD u}ty =@ R}v u Q=ov|oDU@ty
=D
Q iQ u J w Tmar=@ "OyO|t u=Wv =Q Ov=xOW =OH sy R= x t = 12 \ U D w xm (xt xt+12 ) GwR u}@ |]N C@Ft
=@ |ivt x]@=Q xm Ov=xOW =OH sy R= p U = 0.5 x t = 6 CQ Y @ w x Q}O=kt 'CU= |Uwv}U ,=@} QkD |rYi C= Q}}eD
" CU= u=DUtR lJwm Q}O=kt p=@vO x@ OyO|t MQ u =DU@=D QO xm q=@ Q}O=kt p=Ft Q O
|= @ " vQ=O sy
xm |rYi C= Q}}eD w OvwQ P
h L R= TB |v=tR Q
| U QO |oDU@tyOwN X}NWD Q=ov|oDU@ty |rY= O
h y
h L P OvwQ xirw-t w K}LYD |rYi C=Q}}eD decompose() `@=D R= xO =iDU= =@ Q} R O)m QO " CU= "Ov=xOW OQw Q@
"OO oQ |t s}UQD 111 pmW QO Q =ov|oDU@ty w 101 pmW QO |iO=YD xirw-t | U Q TBU "OO o Q |t
> AP.decom <{ decompose(AP, "multiplicative")
> plot(ts(AP.decom$random7:138]))
> acf(AP.decom$random7:138])
|rY Q}}e K}LY C pm | wv}U = rk | Q} x
i C= D D w U= W U m y }= t m OyO|t u=Wv 111 pmW QO Q =ov|oDU@ty
pkDUt w x =
CQ Y @ =Q x t xOR=wO ' y = xirw-t x@ x} RHD =Q} R OUQ|tv CUQO Q_v x@ Q}@aD u}= "CU= xO w@v QFw-t ,qt=m
July R= |rY= =
| yxO=O | U Q Q =}at h= QLv= xm O}vm|t x_Lqt w
'O W s =Hv= | QDW}@ |UQQ@ Qo = "Ovm|t Q
OQw @
C= Q}}eD K}LYD R= TB Q=}at h= QLv= w CU= 41 O vwQ h L R=P TB 'Q=}at h= QLv= "CU= 109 Q Q June =
@= @ D
1.10
1.05
ts(AP.decom$random[7:138])
1.00
0.95
0.90
0 20 40 60 80 100 120
Time
AirPassengers | iO=YD xirw-t %101 pmW
" O}vm xHwD Q} R = O x@ =yQ=}at

| y )
m QLv= x@U=Lt
h= Q CU=
|= @ " 0.03 Q Q
@= @ |rYi

> AP.decom <{ decompose(AP, "multiplicative")
> sd(AP7:138])
1] 109.4187
> sd(AP7:138] - AP.decom$trend7:138])
1] 41.11491
> sd(AP.decom$random7:138])
1] 0.0333884
15 |v=tR = Q
| y| U 1 pYi
Series AP.decom$random[7:138]
1.0
0.8
0.6
0.4
ACF
0.2
0.0
0.2
0 5 10 15 20
Lag
AirPassengers | iO =YD xirw-t Q=ov|oDU@ty %111 pmW

swO pYi
x}=B |iO=YD |=ypOt

|}=yQorta ,=vt "OwW|t EL@ 'OQ=O |Oa@ pwYi QO |O=} R OQ@ Q=m xm |iO=YD |=ypOt R= |=xQ=B pYi u}= QO
p t O QD?U=vt X}NWD Q OO
|= @ | } H Q= @= w R Ovvm|t xO=iDU= |r}N =yv R= |v=tR =
| y| U Q xm Q |t |iQat
OO o
Q |t QDK =w
"OO o w O@=}|t | QDW}@ \U@ R}v xDWPo | y = EL@ R= |=xQ B = u}vJty "OQ}o|t Q=Qk |UQQ@ w
OQ t
1 O}iU xiwv 1 2
xtOkt 1
1
2
Ovm|t u=}@
" t = O
u tR QO =Q p t w x y O =Wt =
| yxO=O | UQ u}@ Cw =iD R= |W=v | =]N 'p tO l} QO x v O =t}k=@
xt |va = O =t}k=@
} yx v Q
| U CQ Y w u}= QO " CU= Zw y
Qit pOt \UwD ^t |va} O
u x W OQw @ Q Q}O=kt w yt Q
| U
"CU= Q} R w x
CQ Y @
xt = yt y^t
;
1. White noise
16
17 x}=B |iO=YD =
| yp t O 2 pYi
Tmavt Q=ov|oDU@ty QO |rYi C= Q}}eD =} w OvwQ Ovv=t |v=tR Q

| U | y = xirw-t xm OW x_Lqt p@k pYi QO
OvW=@ xDU@ty O}=@v =yxOv=t}k=@ Q

| U =t= 'OvDUy xDU@ty |v=tR =
| y| U ,a v Q = w xm OW xO}O u}vJty "O Ww |t
"OO o Q |t Q} R h} QaD ?Hwt xO}= u}= "OyO u=Wv =Q |YNWt |wor= `wv I}y O}=@v =yv Q=ov|oDU@ty =Pr w
h} QaD 2
1
2
w1 w2 wn = Q}eDt Qo = 'Ov} wo 2 (DWN) xDUUo O}iU xiwv =Q f t

| y w : t = 1 2 n g |v=tR |QU
xH}Dv w |t
u= D p qkDU= C}Y=N R= " OvW=@ 2 T = v } Q=w w QiY u}ov=}t =@ u=Um} `} RwD |=Q=O w pkDUt sy R=
w |t x=ov 'Ovm
u= D |w Q}B p=tQv p=tDL= |r=oJ `@=D R= y Cor(wi wj ) = 0 i 6= j x C Q
= Q}eDt Qo = "CU= m i o
" CU= wt N (0 2) x C | = O}i x w Q m W=O Uw o U i v | U
R QO |R U = x}@W 3
1
2
C kw= = |a@ O \UwD xOW

=Q p t |R U = x}@W Q
| U "O W w |t xO =iDU= =yxO=O 3 |R=Ux}@W |= @ Q xDi=} Q
VR= @ p t O
"OO o Q R}=tDt xOW xOy=Wt =
| yxO=O R= =D Ov} wo R}v 4 |oDN=U =
| yxO=O
Q
|= @ u R= u= D w |t w O}=tv O=H}= xOv} Q
|= @ =Q |}=yw} Q=vU Ov=wD|t =F
'p t |= @ " Q CU= | O}it Ov} Qi |R U = x}@W
rnorm(100) =F
p t |= @ " @ } Q O = |t = =
u t U |@ wN@ Q=m u}= R QO "O wtv xO =iDU= O
p t | y = QDt=Q=B u =v}t]= xR @ =
"Ovm O=H}= =Q OvDUy O =D
OQ= v U= p t v u = Q =Um} `} RwD |=Q=O xm pkDUt Q}eDt 100 O v= Dw |t
" O}vm xHwD Q} R = O x@ uwvm =
| y )
m
> set.seed(1)
> w <{ rnorm(100)
> plot.ts(w, type = "l")
= Qy
Q @ QO xm |vat u}O@ 'OW=@ u=Um} |R U = x}@W Q
|= H= Q @ = Qy `w W Q x]kv xm CU= u |= @ Q set.seed() ` = @ D
= |}=D100 hrDNt
| y = Q
| y= H= =@ OwW h L P `@=D u}= Qo = =t= "OO o Q |t O=H}= u =Um} OOa 100 x = Q Q
t v @ |= H=
" CU= 12 pmW q=@ = O

| y )
m |= H= Q xH}Dv Q |t pY=L
"OO o Cw =iDt
Q=ov|oDU@ty w swO x@DQt X=wN 4

1
2
Tv=} Q=wwmwD= `@=D |=Q=O 'OW=@ 2 T = v } Q=w w QiY u}ov=}t |=Q=O w xDU@ty=v w f t g CQwY x@ |iO=YD Q}eDt Qo = w
"CU= Q} R
8
>
< 2 k=0
(k) = >
:0 k=0
6
Q
= }R
Cov(wt wt+k ) = E (wt ; )(wt+k ; )]

2. Discrete white noise 3. Simulation 4. Synthetic
2
1
w
0
1
2
0 20 40 60 80 100
Time
O}iU xiwv |v=tR Q

| U V}=tv %12 pmW
w |t xH}Dv x=ov 'CU=

"O W = E (wt) = E (wt+k ) = 0 w
u J
8
>
E wt = a2 < ( 2) k=0
E (wt wt+k ) = E (wt)E (wt+k ) = >
:0 0 k=0 6
%R= CUDQ=@a |oDU@tyOwN `@=D w
8
>
<1 k=0
k = cck = >
0 :0 k=0
6
x@ pta QO =t= OvW=@ QiY Q@=Q@ O}=@ k=0 6 |=R= x@ xOW = x}@W O}iU xiwv
|R U =
| yxO=O |oDU@tyOwN `@=D
|vat u}O@ u}= w OvwW|t `k=w u=v}t]= xR @ = pN=O =t= 'CU}v QiY ,=k}kO =y k Q}O=kt r |Q}oxvwtv Q}}eD Cra
C=
O
u t uw Q}@ 'Q}N-=D 20 w %5 w
uO @ Q=O |vat K]U x@ xHwD =@ |DL "OvQ=Ov QiY =@ |Q=O|vat Cw=iD =yv xm CU=
"O}vm xHwD Q} R p=Ft x@ uwvm = "CU= p=mW= q@ R}v =y k r R= OQ t w l}
> set.seed(2)
> acf(rnorm(100))
w |a = R
OQ=O O Hw xD@
@ D QO r= "OQ=O |oORuwQ}@ l} 7 Q} =
N-D QO xm 'CU= 22 pmW q=@ = O
| y )
m |= H= Q xH}Dv
C ARMAacf() ` = u
" U= @ D }= " Ovm|t x@U=Lt | Q_v w x
CQ Y @ =Q r
O}iU xiwv l} k |oDU@tyOwN `@=D xm
O}vm xHwD Q} R p=Ft x@
" "O wtv s}UQD plot() ` = =
@ D @ u= D w |t =Q `@=D u}= pY=L
> wn <{ ARMAacf(ar=0, ma=0, lag.max=20)

19 x}=B |iO=YD =
| yp t O 2 pYi
Series rnorm(100)
1.0
0.8
0.6
0.4
ACF
0.2
0.0
0.2
0 5 10 15 20
Lag
O}iU xiwv |oDU@tyOwN `@=D V}=tv %22 pmW
> plot(wn, type='h')

QiY Q@=Q@ Q}O=kt =yQ}N-=D Q}=U Q
|= @ w CU= l} Q@=Q@ k=0 QO xm 'CU= 32 pmW q=@ = O
| y )
m |= H= Q xH}Dv
"CU=
5 |iO=YD uORs=o 2 2
xtOkt 1
2
2
| Q}o =Qi pOt R= u VR= @ Q |DL ,qwtat "CU= |iO=YD = O

| y vwQ |= @ Q |@U=vt Q =@ = |iO=YD
VR= @ , r e uORs o =
" CU= QD?U=vt ARIMA w
u J
h} QaD 2
2
2
Q
% o= CU= |iO=YD =
uORs o f xt g x=ov "CU= |v=tR Q
| U l} f xt g xm O}vm Z i Q
xt = xt 1 + wt ;
=@ u}vJty w q @= xrO=at QO xt 1 = x t 2 + w t
; ; ;1 |v} Ro}=H =@ "CU= O}iU xiwv | U fQ wt x
g u QO m
x
% m O W w |t xH}Dv 6 wQUB |v} Ro}=H VwQ u}= xt=O= w |r@k QO xt 3 |v Ro =
; } } H
xt = wt + wt 1 + wt 2 +
; ;
5. Random walk 6. Back substitution

1.0
0.8
0.6
wn
0.4
0.2
0.0
5 10 15 20
Index
O}iU xiwv | Q_v |oDU@tyOwN `@=D V}=tv %32 pmW
P
%= r "O W w |t Q
`w W t=1 =
u tR R= w CU}v C}=yv|@ w
j i | U Q pta QO
xt = w1 + w2 + + wt
CQ Y @ w x Qorta l} u=wD|t u}=Q@=v@ "OOQo|t p=ta= R}v xO}J}B |v=tR |=y|QU |=ypOt |=Q@ wQUB |v} Ro}=H
"O wtv h} QaD Q} R
7 wQUB p=kDv= Qorta 3

2
2
"CU= Q} R w x xm
CQ Y @ B Qort a
B (xt ) = xt ;1
=@ Q
, D t x WO xDio Qorta Qo = "Ov} wo R}v Q}N-=D Qorta u x@ OQ=wt R= |=xQ=B QO xm 'OwW|t xO}t=v wQUB p=kDv= Qorta
%x =ov Q
'OO o Q=QmD
B n(xt ) = xt ; n
"O W w |t xDWwv Qorta u}= =@ |iO=YD =

uORs o u wvm =
xt = B (xt ) + wt (1 B )xt = wt
) xt = (1 B ) 1 wt
; ) ;
;
= (1 + B + B 2 + )wt xt = wt + wt 1 + wt 2 +
) ; ;
7. Backward shift operator

21 x}=B |iO=YD =
| yp t O 2 pYi
|iO=YD uORs=o swO x@DQt X=wN 4

2
2
" CU= Q} R w x
CQ Y @ x = 0 x @ xHwD =@ |iO=YD uORs o = Ov} Qi swO x@DQt X= N w
t t+k
!
X X X
k (t) = Cov(xt xt+k ) = Cov wi wj = Cov(wi wj )
i=1 j =1 i=j
CU= Q} R
" w x |oDU@tyOwN ?} Q uwvm = "CU= =DU}==v Ov} Qi u}=Q@=v@ 'CU= u=tR
CQ Y @ R= |a@=D Tv=} Q=wwm
k (t) = p Cov(xt xt+k ) = p 2 t 1

2
= p
Var(xt ) Var(xt+k ) t (t + k)2 1 + k=t
l}ORv k Ok u Q =v C QDm w x_ q p = w x t x C@U k Ok = Qy
Q= t }= @ @ " U= J m |= L t @ k CQ Y @ @ v Q= t p L @ nQ @ | yR = t |= @ Q
R= Ov)m |r}N w x
CQ Y @ w O @ w Oy=wN C@Ft Q}O=kt |=Q=O |iO=YD =
uORs o |= @ Q Q =ov|oDU@ty w CU= l} x@
O = |t Vy=m l} Q=Okt
" @ }
8 p=iD Qorta 5
2
2
x
xm OW=@ |iO=YD uORs=o f t g |QU Qo = p=Ft |=Q@ "Ovm p}O@D =DU}= |QU x@ =Q =DU}==v |QU Ov=wD|t Qorta u}=
xt xt 1 = wt
; ; w x
CQ Y @ u x
x]@=Q xm Ovm|t O=H}= =Q O}iU xiwv |QU f t g pw= x@DQt p =iD 'CU= =DU}==v
"CU=
w |t h} QaD Q} R
"O W w x
CQ Y @ r |va} p =iD Qorta
r xt = xt xt ; ; 1
Q =} QUB p=kDv= Qorta ?UL Q@ Ov=wD|t

"OO o u @ w r u}=Q@=v@ "CU= r xt = (1 B )xt x ; m O}W=@ xDW=O xHwD
" CU= Q} R w x p =iD Qorta QDq=@
CQ Y @ | y = x@DQt
r
n
= (1 ; B )n
|R=Ux}@W 6
2
2
"O W w |t Qy=_ =yQ=Owtv QO pOt |rY= C=}YwYN w CU= O}it |v=tR |QU l} xar=]t |=Q@ |R=Ux}@W ,=@r=e
x wkr=@ u=mt= l} u=wvax@ Ov=wD|t pOt =Wv OwN
'O=O u R= =Q x@=Wt C Y =} wYN |N} Q=D xO=O xm |Dkw u}=Q@=v@
"OO oQ |iQat u |= @ Q
" Ovm = x}@W
|R U =Q |iO=YD uORs o = l} Ov=wD|t Q} R | y )
m = O
> set.seed(2)
> x <{ w <{ rnorm(1000)
8. Dierence operator
> for (t in 2:1000) xt] <{ xt - 1] + wt]

> plot(x, type = "l")
= for xkr
uORs o Q| | Ok x x =vt
L "OO o t yOQ= t @ , w OyO|t C@Uv wx @ =Q O}iU xiwv Q
| U Q wDUO u}rw=
xkrL R= u QO xm Owtv O=H}= |}=yO)m =@ =Q |iO=YD =
uORs o |R U = x}@W w |t xD@r= "Ovm|t O=H}=
u= D =Q |iO=YD
O = OWv
" W @ x xO=iDU=
> set.seed(2) # so you can reproduce the results

> v <{ rnorm(1000) # v contains 1000 iid N(0,1) variates
> x <{ cumsum(v) # x is a random walk
> plot(x, type = "l")
Q
| U Q =ov|oDU@ty uOQw CUOx@ Q CU= xOW
|= @ " xO=O u =Wv 42 pmW O
QO x W O =H}= Q
| U |wQ Qy@
60
40
x
20
0
0 200 400 600 800 1000
Index
|iO=YD = V}=tv
uORs o %42 pmW
"O Ww |t xOy=Wt 52 pmW QO xH}Dv TBU "OOQo|t =QH= Q} R QwDUO 'p@k |=yxt=vQ@ xt=O= QO 'xOW |R=Ux}@W
> acf(x)
"OO o Q |t O}iU xiwv =
| yxO=O | U Q x@ p}O@D p =iD l} QO |iO=YD = O
uORs o ' W x_Lqt ,q@k xm Qw]u=ty
p =iD Qorta p=ta=
' Q CU= jO=Y R}v xOW
|= @ " = x}@W
|R U Q w
| U OQ t QO x W O xDio ?r]t xm O}O O}=@ uwvm =
62 pmW QO xH}Dv TBU w |t =QH= Q} R QwDUO 'Q}N=
"O W = x = Q xt=O=
| y t v @ QO " CU= OwHwt R di() ` =
QO @ D
"O W w |t xOy=Wt
> acf(di(x))
u =v}t]= =
xR @ R= |oORuwQ}@ w
OQ t wO "O W w |tv x_Lqt Q=ov|oDU@ty Q |YNWt
|= @ |wor= 'Q}N= pmW QO
|R U = x}@W |QU CiQ|t Q=_Dv= xm Qw]uty u}=Q@=v@ "CU}v =vDa= p@=k %5 uOw@ Q=O|vat K]U QO xm OQ=O OwHw
23 x}=B |iO=YD =
| yp t O 2 pYi
Series x
1.0
0.8
0.6
ACF
0.4
0.2
0.0
0 5 10 15 20 25 30
Lag
|iO=YD uORs o Q = =ov|oDU@tyOwN V}=tv %52 pmW
"Ovm|t |wQ}B |iO=YD uORs o = Ov} Qi l} R= x x W O
w}UQoQwD= |=ypOt 3 2
h} QaD 1
3
2
Qo = 'Ov} wo AR(p) Q=YDN= x@ p x@ QD t w}UQoQwD= Ov} Qi =Q f xt g |v=tR | U Q
xt = 1 xt 1 + 2xt 2 +
; ; + pxt p + wt ;
p=kDv= Qorta ?UL Q@ Q}N= xrO=at Qo = "OvDUy p = 0 =6 O

@ p t | y = QDt=Q=B i w O}iU xiwv f wt g u QO xm
"O @w Oy=wN Q} R w x
CQ Y @ 'OO o u @ w Q =} QUB

p(B )xt = (1 1B 2B 2 ; ; ; ; pB p)xt = wt
"O}vm xHwD Q} R C =mv x@ uwvm =
O = |t
" W @ 1 = 1 u QO xm CU= AR(1) X N = Cr=L |iO=YD uORs o =
w}UQoQwD= u x@ ?@U u}ty x@ w CU= xDWPo =

| yu tR QO= x x
=y x@ C@Uv t uw}UQoQ `k=w QO p t O u}=
"Ov} wo
% @ = CU= Q@=Q@ t =
u tR QO OQw @ Q
x^t = 1xt 1 + 2xt 2 +

; ; + t pxt
;
Series diff(x)
1.0
0.8
0.6
ACF
0.4
0.2
0.0
0 5 10 15 20 25 30
Lag
p=iD p=ta= =@ |iO=YD =

uORs o Q =ov|oDU@tyOwN V}=tv %62 pmW
"OO oQ |t OQw @Q =]N C =a@ Qt pk=OL VwQ x@ pOt = QDt=Q=B

| y
w}UQoQwD= QO |}=DU}==v w |}=DU}= 2

3
2
Q k O Qo = "OW=@ \rDNt =} w |k}kL = xW} Q

| y |=Q=O Ov=wD|t xm Ovt=v xYNWt xrO=at =Q
p(B ) = 0 x rO =at
uORs o = Ov} Qi QO " CU= =DU}= Ov} Qi x=ov 'OO oQ OL=w R= QDW}@ p (
B ) xrt Ov = xW =t jr] |= H J | y }Q s D t
u}}aD Q Q} R p=Ft Q=yJ x@ uwvm = "CU= =DU}==v Ov} Qi

|= @ w O W w | B = 1 x C
= 1 B | =Y
t m U= ; iO D
O}v x w AR Ov Q | =DU = | =DU

" m H D } i } }= v w } }=
B=2QQ 1 @= @ ;
1
2 B=0x rO =at xW} Q =Q} R 'CU= =DU}= t x = 12 xt 1 + wt ; CQ Y @w x AR(1) O
p t 1
" CU= QDoQR@ OL=w R= xm CU=
w QUB p=kDv= Qorta R= xO =iDU= =@ =Q} R "CU= =DU}= t x = xt 1 41 xt 2 + wt w x AR(2) O

; ; ; CQ Y @ p t 2
4 (B 2)2 xt = wt |va 41 (B 2 4B + 4)xt = wt x =a

1
|= xrtHOvJ u}= xW} Q "OwW|t xH}Dv ; } ; rO t
O = |t OL=w
" W @ R= QDoQR@ w CU= B = 2 x O | C x
(B ) = 14 (B 2)2 x =a p =
m } t UO @ ; rO t L @
" CU= OL=w =yxW} Q x = 21 xt 1 + 12 xt 2 + wt w x AR(2) O

R= |m} =Q} R 'CU= =DU}==v t ; ; CQ Y @ p t 3
2 (B 1)(B + 2)xt = wt |va 12 (B 2 + B 2)xt = wt QU =kD Qort

1
; ; =iD =
} ; ; w B p v= a R= xO U= @
C B = 1 2 = xW
" U=
(B ) = 21 (B 1)(B + 2) xrt Ov w | xH}D
; | y } Q |=Q=O ; ; |= H J "O W t v
C O "QD R B = 2 jr] O Q
U= L=w R= Q | | =DU = ?@ B = 1 Ok \k
oQ @ ; t Q k = } R "OO o t } }= v U Q= t i
|va} 1 + 41 B 2 = 0 | y = xW} Q =Q} R 'CU= =DU}= xt = ;

1
x
4 t;2 + wt CQ Y @ w x AR(2) O
p t 4
25 x}=B |iO=YD =
| yp t O 2 pYi
i= B = 2i
p
2QQ = xW} Q
@= @ y R= l} Qy jr]t QOk "OW=@|t ; 1 w CU= \rDNt O
O= a= xm OvDUy
"CU= OL=w R= QDoQR@ xm OW=@|t
xrO=at | y = xW} Q w |t xm
u= D OQ=O O Hw w polyroot() = x@ |a@=D
s v R QO |= xrtHOvJ = xW} Q x@U=Lt
| y Q
|= @
O
x W xDio `@=D =@ sQ =yJ w w
w s U OQ= t 'p t u= =F wvax@ Q |UQQ@
"O m =Q =y|QU |}=DU}= w O wtv x@U=Lt =Q xYNWt
w |t x@U=Lt
"O W
w
%s U p t =F
> polyroot(c(-2,1,1))
1] 1-0i -2+0i
%R= OvDQ=@a =yxW} Q jr]t QOk
> Mod(polyroot(c(-2,1,1)))
1] 1 2
%sQ =yJ p=Ft
> polyroot(c(1,0,1/4))
1] 0+2i 0-2i
%R= OvDQ=@a =yxW} Q jr]t QOk
> Mod(polyroot(c(1,0,1/4)))
1] 2 2
AR(1) p t swO O x@DQt X=wN 3

3
2
" CU= Q} R CQ Y @w x AR(1) O O

p t ' W x y O =Wt ,q@k xm Qw]u=ty
xt = xt 1 + wt ;
xm O=O u =Wv u= Dw |t "OW=@|t 2 T = v } Q=w w QiY u}ov=}t =@ O}iU xiwv f wt g u QO xm
x = 0
k = k 2=(1 2) ;
xm CWwv w |t
u= D j < 1 Ov Q
j } i |}=DU}= Q
\ W w B Qort a R= xO =iDU= =@
(1 ; B )xt = wt
"O wtv x@U=Lt =Q xt u= Dw |t uwvm =
xt = (1 B ) 1wt;
;
X
1
= wt + wt 1 + wt 2 +
;
2
; = i wt ;i
i=0
" O}vm xHwD u}ov=}t x@U=Lt x@ uwvm =

!
X
1
X
1
E (xt ) = E i wt ; i = iE (wt i) = 0 ;
i=0 i=0
"CU= Q} R KQW x@ Tv=} Q=wwmwD=

1 !
X X
1
k = Cov(xt xt+k ) = Cov wt i

i
; j wt+k ; j
i=0 j =0
X
= j =k +i
i j Cov(wt i wt+k j ) ; ;
X1
= k 2 i=0
2i = k 2=(1 2) ;
AR(1) Ov}Qi Q=ov|oDU@ty 4

3
2
"O}vm s}UkD Tv=} Q=w Q@ =Q Tv=} Q=wwmwD= xm CU= |i=m |oDU@tyOwN `@=D x@U=Lt |= @Q
k = k (k
0)
p}t QiY CtU x@ | QDW}@ CaQU =@ QDmJwm | y = |=R= x@ Q=ov|oDU@ty u}=Q@=v@ "CU= j <1
j u QO xm
O |t
" } 72 pmW QO |iv t w C@Ft Q}O=kt Q =ov|oDU@ty
|= @ Q wO p t =F xt=O= QO " Ovm|t
> plot(0:10, rho(0:10, -0.7), type = "b", xlab="lag k")

> rho <{ function(k, alpha) alpha^k
> par(mfrow=c(2,1), mar=c(4,4,2,4))
> plot(0:10, rho(0:10, 0.7), type = "b", xlab="lag k",
+ ylab=expression(rk]), main=expression(alpha==0.7))
> plot(0:10, rho(0:10, -0.7), type = "b", xlab="lag k",
+ ylab=expression(rk]), main=expression(alpha==-0.7))
> abline(h=0, lty=2)
9 |} RH |oDU@tyOwN `@=D 5
3
2
x
x@ \ki t Qo = |DL 'OvDUy QiY hr=Nt =yQ}N-=D s=tD |= @ Q =y|oDU@tyOwN Q}O=kt ' k Q} N= xrO=at j@=]t
k
=y|oDU@ty QF= xm CU= |oDU@ty R= |awv ' Q}N-=D QO |} RH |oDU@tyOwN `@=D "OW=@ xDW=O |oDU@ t;1 x
"O @w Oy=wN QiY k>1 s =tD |= @QAR(1) | R } H |oDU@tyOwN `@=D p=Ft Q Q
|= @ "OO o h L P QDmJwm Q}N-=D =@
CU=
' AR(k) O xDi=}
p t VR= @ Q ? Q u} k =
} t= @ Q@=Q@ k Q} = N-D QO |} RH |oDU@tyOwN `@=D '|rm Cr=L QO
|oDU@tyOwN `@=D Q=ov|oDU@ty u}=Q@=v@ "CU= QiY QQ k>p

@= @ |=R= x@ k ?}=Q AR(p) Ov} Qi QO |va}
w x@U=Lt Q
|= @ pacf() = x@ |a@=D
s v R QO "O Q}o Q=Qk xO =iDU= w
OQ t O w | AR Ov Q
v= D t } i x@DQt u}}aD Q |} RH
|= @
"OQ=O O Hw w |} RH |oDU@tyOwN `@=D s}UQD

9. Partial autocorrelation function
27 x}=B |iO=YD =
| yp t O 2 pYi
= 0.7
0.8
rk
0.4
0.0
0 2 4 6 8 10
lag k
= 0.7
1.0
0.5
rk
0.0
0.5
0 2 4 6 8 10
lag k
= 0:7 ;0:7 Q
|= @ AR(1) Ov Q } i Q =ov|oDU@ty V}=tv %72 pmW
|R=Ux}@W 6
3
2
QO |} RH |oDU@tyOwN w |oDU@tyOwN `@=wD w Ov} Qi O

w x W |R U = x}@W Q} R CQ Y @w x R QO AR(1) O
p t
|oDU@tyOwN Q=ov|oDU@ty QO k>1Q }O=kt |= @Q O}vm|t x_Lqt xm Qw]u=ty "OOQo|t s}UQD 82 pmW
O w
"OQ= v O Hw |Q=O |vat |oDU@ty '|} RH
xDi=} VR=Q@ |=ypOt 7

3
2
xOW |R=Ux}@W |=yxO=O Q@ xDi=} VR=Q@ |=ypOt 8

3
2
Q
| U |= @ Q x.ar uw}UQoQwD= pOt 'Q} R |=yO)m O = |t VR=Q@ =yxO=O Q@ ar() `@=D \UwD R QO
QO " @ } AR(p) O
p t
R= xO =iDU= =@ u
10 |@v=Ht Tv=} Q=w xm 'O@=}|t Q O
VR= @ x W xO=O QDt=Q=B w %95 u=v}t]= = = O
xR @ @ x W |R U = x}@W
"O Ww |t G=QNDU= x.ar$asy.var
> set.seed(1)
> x <{ w <{ rnorm(100)
> for (t in 2:100) xt] <- 0.7 xt - 1] + wt]
> x.ar <{ ar(x, method = "mle")
> x.ar$order
pOt x@DQ
1] 1
> x.ar$ar
pOt QDt=Q=B
10. asymtotic
3
2
1
x
0
1
0 20 40 60 80 100
Index
1.0
0.6
ACF
0.2
0.2
0 5 10 15 20
Lag
0.2 0.4 0.6
Partial ACF
0.2
5 10 15 20
Lag
V}=y|oDU@tyOwN w AR(1) Ov Q } i V}=tv %82 pmW
1] 0.6009459
> x.ar$ar + c(-2, 2) sqrt(x.ar$asy.var)
pOt QDt=Q=B u=v}t]= xR=@
1] 0.4404031 0.7614886

? =NDv= "CU= Q=wDU= |}=tvDUQO QFm =OL `@=D T U= = Q@ 'OW xO =iDU= =yO)m QO VR= @ Q Ov} Qi QO xm 11 mle VwQ
x]@= x@U=Lt "CU= pOt QDt=Q=B pk=OL Qov=}@ xm OwW|t s=Hv= 12 AIC x]@= R= 'Ov} Qi x@ \w@ Qt p Okt
Q=
" CU= Q} R w x O
CQ Y @ x W xDio
AIC = ;2 log-likelihood + 2 number of parameters

CUQO x@DQt =
'q @ | y )
m= O xm O}vm xHwD "Ovm|t Q=}DN= =Q VR= @ Q u} QDy@ AIC u QDm } w =
J m @ p t ' Oar() ` = @ D QO
Q= Okt R= QDmJwm xm 'Ot CUOx@ ^ = 0:6 Q Q AR(1) @= @ O QDt=Q=B

p t OQw @ "OQ wtv |@=} R=@ =Q p = 1 |va O } p t
O
"OQ= v O Hw w E}L u}= Q
R= |O= }= w OO o Q |t pOt QDt=Q=B Q=Okt pt=W %95 u=v}t]= xR @ = =t= C = 0:7
" U=
xDi=} VR=Q@ AR p t % O |v=yH |=tO |QU 9

3
2
=aO= u= Dw |t O}OQD =@ "CU= |= v x =Nro QF= R= |W=v xm OyO|t u=Wv =Q tO = OvwQ V}=Ri= 1970 =
p U R= | U Q u}=
| U Q =@ =aO= u}= = =
|Q oR U |= @ " Q CU= xOW h L P |iO=YD xO}OB l} p=kDv= R= |W=v |=x Ov}=Ri OvwQ xm Owtv
11. Maximum likelihood estimation 12. Akaike Information Criterion
29 x}=B |iO=YD =
| yp tO 2 pYi
"O wtv xO =iDU= u}at `@=wD R= xm O

u uw @ xO wtv =
|R Up t =Q O OvwQ xirw-t CU= umtt '=yxO=O
"O}vm xHwD xvq=U =
| tO | U Q u}ov=}t x@ xDi=} VR= @ Q AR p t O x@
> www <{ "http://www.massey.ac.nz/~pscowper/ts/global.dat"

> Global <{ scan(www)
> Global.ts <{ ts(Global, st = c(1856, 1), end = c(2005, 12), fr = 12)
> Global.ar <{ ar(aggregate(Global.ts, FUN = mean), method = "mle")
> mean(aggregate(Global.ts, FUN = mean))
1] -0.1382628
> Global.ar$order
1] 4
> Global.ar$ar
1] 0.58762026 0.01260253 0.11116731 0.26763656
> acf(Global.ar$res-(1:Global.ar$order)], lag = 50, main="")
1.0
0.8
0.6
ACF
0.4
0.2
0.0
0.2
0 10 20 30 40 50
Lag
=yxOv=t}k=@ Q=ov|oDU@ty V}=tv %92 pmW
Q}N-=D QO rk Q=Okt RH x@ =Q} R "OyO|t u=Wv =Q O}iU xiwv Q

| U l} '=yxOv=t}k=@ 92 pmW Q=ov|oDU@ty
O
" UQ |t Q_v x@ ?U=vt '=yxO=O Q@ AR(4) O Q P O
p t VR= @ = r " vQ=O Q= k u Q =v}t]= =
xR @ QO =yv x}k@ 27
xO=O CUOx@ O
=Q p t | y = QDt=Q=B w xvq=U \UwDt =
| tO u}ov=}t Q=Okt xm Q}N= = O G}=Dv x@ xHwD =@ uwvm =
| y )
m
CWwv
" u= Dw |t 'CU=
x^t = 0:14+0:59(xt 1 +0:14)+0:013(xt 2+0:14)+0:11(xt 3 +0:15)+0:27(xt 4 +0:15)

; ; ; ; ;
swU pYi
=DU}= |=ypOt
O}m =
,= xt
=Q f g |v=tR Q
| U "O W w |t | Q}o}B EL@ u xt=O= u wvm = "OW KQ]t |@r=]t |}=DU}= x@ `H=Q ,q@k
|va} 'Ovmv Q}}eD u=tR QO p =kDv= QF= Q
QO | U s= D w `} RwD `@=D Qo = Ov} wo 1 =DU}=
f (xt1 xt2 xt ) = f (xt1 +h xt2 +h

k
xt +h)
k
Cov(xt xs) T = w
v } Q=w m w O W w |t pt=W R}v =Q u=tR QO C@=F Tv=} Q=w w u}ov=}t 'Omw-t |}=DU}= xm O}vm xHwD
O = xDW=O |oDU@
" W @ k= t s
j ; j Q}N-=D x@ \ki
lQLDt u}ov=}t |=ypOt 1 3
X=wN w h} QaD MA(q)

% Ov}Qi 1
1
3
xrtH quQ } N u}vJty w O}iU xiwv |Q H = xrtH R= |]N ?}mQD q x@ QD t R= (MA) l QLDt u}ov=}t Ov} Qi
" CU= Q} R w x
CQ Y @ u x]@=Q "OW=@|t O}iU xiwv |r@k
xt = wt + 1 wt 1 + ; + q wt ; q
1. Strictly stationary
30
31 =DU}= | yp t= O 3 pYi
p=kDv= Qorta ?UL Q@ w |t

u= D =Q q @
2
= xrO=at "CU= w Tv=} Q=w w QiY u}ov=}t =@ O}iU xiwv f wt g u QO xm
"O wtv |U} wvR=@ B
' w QUB
xt = (1 + 1B + 2B 2 + + q B q )wt = q (B )wt
| =DU}= CqtH R= |y=vDt `tH pt=W MA |=yOv} Qi xm u}= x@ Q_v "CU= qx HQO R= |= xrtHOvJ q u QO xm
w
"OQ=O O Hw u tR = x@ C@Uv Tv=} Q=wwmwD= w u}ov=}t QO C F w =@ CU= Q=QkQ@ |}=DU}= =Pr 'CU= O}iU xiwv
" CU= x@U=Lt p@=k |oO=U x@ fxt g Tv=} Q=w w u}ov=}t
E (xt ) = E (wt + 1 wt 1 + + q wt q ) ; ;
= E (wt) + 1 E (wt 1) + + q E (wt q ) ; ;
=0
" OvDUy QiY Q@=Q@ =yu}ov=}t R= l} Qy =Q} R
Var(xt ) = Var(wt + 1wt 1 + + q wt q ) ; ;
= Var(wt) + 12 Var(wt 1) + + q2 Var(wt q ) ; ;
= w2 (1 + 12 + + q2)
" OvDUy pkDUt Qo}Om} R= w w2 u =Um} Tv=} Q=w |=Q=O C qtH s=tD =Q} R
" CU= Q} R QwYx@ k 0
|= @Q |oDU@tyOwN `@=D
8
>
>
>
>
>
1 k=0
P >
> q ;k
< i i+k
(k) = > P 2 i=0
q k = 1 : : : q
>
> i
>
> i=0
>
>
: 0 k>q
O}iU xiwv pkDUt C qtH pt=W t;1x w xt Q
= }R ' CU= QiY Q@=Q@ `@=D k>q Q CU=
|= @ " 0 = 1 u QO xm
" CU= QiY =yv Tv=} Q=wwm u}vJty w OvDUy
xrtH uw @ O |y=vDt=v x@DQt =@ =DU}= w}UQoQwD= Ov} Qi ?UL Q@ =Q u u= wD@ Qo = Ov} wo 2 Q}PBuwQ=w =Q MA Ov Q } i
w
"O W u @ =} Q} R w x Ov=wD|t t
CQ Y @ x = (1 B )wt x] ; @=Q =@ MA(1) Ov Q =F
} i 'p t |= @ " Q CWwv =]N
wt = (1 B ) 1xt = xt + xt 1 + 2xt 2 +
;
;
; ;
"Ovm O
j Y |}=Qoty \ W Q =D CU= j <1 j u QO xm
OL=w R= QDoQR@ |oty (B ) = xW} Q jr]t QOk Qo = CU= Q}PBuwQ=w Ov} Qi l}
| y MA(q) Ov Q } i ' |rm Qw]x@
" OvW=@
2. invertible
|R=Ux}@W w Q=ov|oDU@ty % R | yp = =Ft 2

1
3
w xkrL R= xO=iDU= =@ |a@=D Q=m u}= Q

|= @ "O wtv =}
xO B R QO u= D w |t =Q MA(q) Ov Q Q |oDU@tyOwN `@=D
} i |= @
"O W w |t h} QaD Q
\ W
> rho <{ function(k, beta) f

+ q <{ length(beta) - 1
+ if (k > q) ACF <{ 0 else f
+ s1 <{ 0 s2 <{ 0
+ for (i in 1:(q-k+1)) s1 <- s1 + betai] betai+k]
+ for (i in 1:(q+1)) s2 <{ s2 + betai]^2
+ ACF <{ s1 / s2g
+ ACFg
Ov Q Q Q
} i |= @ = O C x@ =L p = MA(q) Ov Q
}R | y )
m " U= U t Q |oDU@t w ` = = = O
@ k =iD =
} i |= @ yO N @ D 'q @ | y )
m R= xO U= @
C " pm w x 3 = 0:2 = 0:5 1 = 0:7 = QD = = MA(3)

U= 13 W CQ Y @ w ' | y t=Q B @
> beta <{ c(1, 0.7, 0.5, 0.2)

> rho.k <{ rep(1, 10)
> for (k in 1:10) rho.kk] <{ rho(k, beta)
> plot(0:10, c(1, rho.k), pch = 4, ylab = expression(rhok]), xlab="lag k")
> abline(0, 0)
1.0
0.8
0.6
k
0.4
0.2
0.0
0 2 4 6 8 10
lag k
MA(3) O
p t Q =ov|oDU@ty V}=tv %13 pmW
33 =DU}= =
| yp tO 3 pYi
w |v=tR Q
| U "O Q}o|t Q=Qk =iDU=
xO w
OQ t Q =ov|oDU@ty s}UQD w MA(3) Ov Q = x}@W
} i |R U Q Q} R
|= @ = O
| y )
m
|oDU@tyOwN `@=D Q=Okt xU u}rw= 'CiQ|t Q=_Dv= xm Qw]u=ty "CU= xOW s}UQD 23 pmW QO Q=ov|oDU@ty
O Q
" vQ=O Q= k u =v}t]= =
xR @ QO r
=y k Q}O=kt '3 R= QDW}@ | y = Q}N-=D Q
|= @ w OvDUy QiY R= |Q=O |vat Cw =iD |=Q=O
"CU= | Q}oxvwtv Q}}eD

C= R= |W=v xm OW=@ xDW=O OwHw |oORuwQ}@ CU= umtt sy %5 xD@r=
> set.seed(1)
> b <{ c(0.8, 0.6, 0.4)
> x <{ w <{ rnorm(1000)
> for (t in 4:1000) f
+ for (j in 1:3) xt] <{ xt] + bj] wt - j]
+g
> par(mfrow=c(2,1), mar=c(4,4,2,4))
> plot(x, type = "l", xlab="time t")
> acf(x, main="")
4
2
x
0
2
4
0 200 400 600 800 1000
time t
0.8
ACF
0.4
0.0
0 5 10 15 20 25 30
Lag
u Q =ov|oDU@ty w MA(3) |R U = x}@W Ov}Qi V}=tv %23 pmW

xDi=} VR=Q@ MA |=ypOt 2 3
xOW |R=Ux}@W |QU x@ xDi=} VR=Q@ pOt 1

2
3
= w
u t oQ =@ pOt x@DQt xm w
OQ=O O Hw arima() = x@ |a@=D
s v R O= Q =
QO " @ } VR= @ yxO=O Q@ Ov=wD|t MA(q) O p t
QUm =Q u}ov=}t Z i Q V}B w x

CQ Y @ x W O xDio `@=D ' ar() ` = @ D h qN Q@ "O Ww |t s}_vD order=c(0,0,q)
"OO oQ |t OQw @Q 3 <=O@t R= Z a Q xrtH QO w Ovm|tv
C =a@ Qt `wtHt '=yQDt=Q=B OQw Q@ |=Q@ arima() `@=D "OvwW|t OQw Q@ |OOa sD} Qwor= l} \UwD pOt |=yQDt=Q=B
O =
" W @ method=c("CSS") = w x | w u t oQ m DQ Y QO xD@r= OR U = |t pk=OL =Q |]QW
= QDt=Q=B
' y |=R= x@ xm CU= CQ Y w u x MA(q) Ov Q
}= @ Q Q | Q } i VR= @ |= @ ] W C =a@ Qt `wtHt sD} Qwor= K} wD
Q} R CQ Y @ w x w^t |va } VOQw @ wQ wt = O =t} = Q | x@ =L Qm
yx v k @ "OO o t U t Q w x =yxOv=t}k=@
t Q ] @ C =a@ Qt `wtHt
"CU=
n
X n n
X o2
S (^1 : : : ^q ) = w^t =
2
xt ; (^1w^t 1 + ; + ^q w^t q ) ;
t=1 t=1
R= | }O Q =kt '|O a |O wHDUH "OvW=@ QiY =@ Q@=Q@ '=yv Q=QmD Q

`w W QO w^0 : : : w^t q Q}O=kt xm
; u Q@ \w QWt
"OO oQ pk=OL q=@ C =a@ Qt `wtHt xm Ovm|t u}at =Q =yQDt=Q=B
| y = QDt=Q=B "O@=}|t VR= @ Q Q}N= xOW |R U = x}@W =
| yxO=O x@x.ma 'l QLDt u}ov=}t O Q
p t ' }R | y )
m QO = O
pt=W coeff: 2 s.e. of coeff: =@ Qk
, } D O Q}o|t Q=Qk %95 =v}t u =
]= xR @ QO xt=vQ@ |HwQN O
QO x W OQw @ Q
Q =_Dv= xm u=vJty ,=vt "CU= xOW xO=iDU= =yv R= |R=Ux}@W QO xm OwW|t (0.8,0.6,0.4) |=yQDt=Q=B Q}O=kt
"OQ= v O QiY =@ |Q=O |vat Cw =iD <= O@t R= Z a Q CiQ|t
`@=D include.mean=FALSE = w
QO = u Q u t oQ R= Q m }= |= @ "O wtv s}_vD QiY Q@=Q@ =Q u}ov=}t Q=Okt u= D w |t
C include.mean=TRUE
" U= Z i Q V}B u}=Q@=v@ w |t
"O W xO =iDU= arima()
> set.seed(1)
> b <{ c(0.8, 0.6, 0.4)
> x <{ w <{ rnorm(1000)
> for (t in 4:1000) f
+ for (j in 1:3) xt] <{ xt] + bj] wt - j]
+g
> x.ma <{ arima(x, order = c(0, 0, 3))
> x.ma
Call:
arima(x = x, order = c(0, 0, 3))
3. Intercept
35 =DU}= =
| yp t O 3 pYi
Coefficients:
ma1 ma2 ma3 intercept
0.7898 0.5665 0.3959 -0.0322
s.e. 0.0307 0.0351 0.0320 0.0898
sigma^2 estimated as 1.068: log likelihood = -1452.41, aic = 2914.83
ARMA Ov}Qi %|@}mQD |=ypOt 3 3
h} QaD 1
3
3
" CU= Q} R CQ Y @w x p x@ Q D t R= w}UQoQwD= Ov} Qi f xt g |QU 'OW xOy=Wt ,q@k xm Qw]u=ty
xt = 1 xt 1 + 2xt 2 +
; ; + pxt p + wt ;
MA AR Ov Q x |D = O
w } i m kw yp t R= | O}it xDUO "OvDUy pOt | y = QDt=Q=B =y i w O}iU xiwv f t g w u QO xm
(ARMA) QLD u}o =} w} Q
l t v t U oQ D=w Ov} Qi |=Q=O f xt g |v=tR |QU "OO o Q |t pY=L 'OvwW|t xOwRi= sy x@
"CU= Q} R w x
CQ Y @ u x]@=Q w |t
"O W xO=O u =Wv ARMA(p,q) = @ w CU= (p,q) x@ Q D t R=
xt = 1xt 1 + 2xt 2 +
; ; + pxt p + wt + 1wt 1 + 2 wt 2 +
; ; ; + q wt ; q
" CWwv wQUB p=kDv= Qorta w x

CQ Y @ u= D w |t =Q j i w x]@=Q "CU= O}iU xiwv f wt g u QO xm

p(B )xt = q (B )wt
" CU= xHwD QwNQO ARMA(p,q) w

OQ t QO Q} R w
OQ= t
" OvW=@ OL=w R= QDoQR@

| y = xW} Q s=tD jr]t QOk Qo = CU= =DU}= Ov} Qi 1
"OvW=@ OL=w R= QDoQR@ | y = xW} Q s=tD jr]t QOk Qo = CU= Q}PBuwQ=w Ov} Qi 2
" CU= ARMA(p,0) O

p t X N = Cr=L AR(p) O
p t 3
" CU= ARMA(0,q) O

p t X N = Cr=L MA(q) O
p t 4
=yQDt=Q=B X wYN QO | QDW}@ |} Q=m ,=@r=e ARMA O

p t 'O W w |t s=Hv= VR= @ Q |Dkw %QDt=Q=B pk=OL 5
"OQ=O |}=yvD x@ AR = MA } w =
| yp t O x@ C@Uv
pOt swO x@DQt X=wN 2

3
3
C qtH ?UL Q@ =Q u =OD@= QO xm CU= QDy@ 'f t g |v=tR x | U Q ARMA(p,q) swO x@DQt X=wN u}}aD |=Q@
O Q}o@ Q_v
" } QO =Q ARMA(1,1) O =F
p t p t |= @ " Q OvDUy pkDUt sy R= =yv =Q} R "O}U} wv@ O}iU xiwv
xt = xt 1 + wt + wt ; ; 1
Q@ =Q Q}N= xrO=at "CU= Var(wt) = w2 E (wt ) w u Tv=} Q=w w u}ov=}t "CU= O}iU xiwv wt u QO xm
O
" } QO O}iU xiwv | y = xirw-t ?UL Q@ xm xO wtv ?DQt w
|Q ] xt =
T U=
xt = (1 B ) 1(1 + B )wt
;
;
"CU= Q} R w x
CQ Y @ q @ = x]@=Q CU=Q Q \U@
h ]
xt = (1 + B + 2B 2 + )(1 + B )wt
!
X1
= iB i (1 + B )wt
i=0
!
X
1
X
1
= 1+ i+1B i+1 + iB i+1 wt

i=0 i=0
X
1
= wt + ( + ) i 1wt
;
; i
i=0
Tv=} Q=w w CU= E (wt i) = 0 = i

; y s =tD |=R= x@ =Q} R 'CU= QiY Q@=Q@ E (xt ) u}o =} v t 'q @ = x]@=Q x@ xHwD =@
" #
X
1
Var(xt ) = Var wt + ( + ) i 1wt

;
; i
i=1
= w2 + w2 ( + )(1 + 2) ; 1
w |t xH}Dv
"O W k>0 Q
|= @ k T = ww
v } Q=w m D= |= @Q
X
1
Cov(xt xt+k ) = ( + ) k 1w2 + ( + )2w2 k

;
2i ;2
i=0
= ( + ) k 1w2 + ( + )2w2 k (1 ; 2)
; ; 1
"O W
w |t xH}Dv k |oDU@tyOwN `@=D |= @ Q
k = k =0 = Cov(xt xt+k )= Var(xt )

= ( + )(1 +2 )
k 1 ;
1 + +
37 =DU}= =
| yp t O 3 pYi
k = k ; 1 CiQo xH}Dv w |t Q}N= x]@=Q

u= D R=
|@ QHDp}rLD %ARMA |=ypOt 4 3
VR=Q@ w |R=Ux}@W 1
4
3
`@=D \UwD w |t
u= D w |t EL@
=Q O W | Oa@ pYi QO ARIMA
xm = Ov} Qi
| y u R= QD|twta ARMA Ov Q
w } i
c(p,0,q) x@ Q D t @= arima() ` = @ D \UwD Ov=wD|t ARMA(p,q) O

p t "O wtv |R U = x}@W R arima.sim()
QO
Q
"O=O VR= @
O TBU
p t w |t
'O W |R U= x}@W = 0:5 = 0:6 = ARMA(1,1) Ov Q
w ; @ Q =
} i |= @ yxO=O Q} R = O
| y )
m QO
l}ORv xvwtv R= w = QDt=Q=B

| y Q CiQ|t Q=_Dv= xm Qw]u=ty "O@=}|t
OQw @ VR= @ Q =yv Q@ ARMA(1,1)
"CU= q=@ QO =yv Q}O=kt
> set.seed(1)
> x <{ arima.sim(n = 10000, list(ar = -0.6, ma = 0.5))
> coef(arima(x, order = c(1, 0, 1)))
ar1 ma1 intercept

-0.596966371 0.502703368 -0.006571345
p}O@D MQv |QU 2

4
3
x]@= w O@=}|t VR= @Q p}O@D Q

Mv | U Q Q@ARMA(1,1) AR(1) MA(1) w ' = O Q
| yp t ' } R | y )
m QO = O
33 pmW "OW=@|t =yxO=O Q
|= @ | D Q ?U =v O ARMA(1,1) Ov Q O Q |
t p t } i " vO o t xU}=kt sy =@ =yv AIC
CU= lJwm
w ' | y= |oDU@tyOwN |=Q=O x O | =W
m ARMA(1,1) O
yO t u v =Q p t x v O =t}k=@ Q=ov|oDU@ty
" Ovm|t O}}=D =Q p tO u}= R= xO =iDU= w OvDUy O}iU xiwv =@ Q=oR=U
> www <{ "http://www.massey.ac.nz/~pscowper/ts/pounds_nz.dat"

> x <{ read.table(www, header = T)
> x.ts <{ ts(x, st = 1991, fr = 4)
> x.ma <{ arima(x.ts, order = c(0, 0, 1))
> x.ar <{ arima(x.ts, order = c(1, 0, 0))
> x.arma <{ arima(x.ts, order = c(1, 0, 1))
> AIC(x.ma)
1] -3.526895
> AIC(x.ar)
1] -37.40417
> AIC(x.arma)
1] -42.27357
> x.arma
Call:
arima(x = x.ts, order = c(1, 0, 1))
Coefficients:
ar1 ma1 intercept
0.892 0.532 2.960
s.e. 0.076 0.202 0.244
sigma^2 estimated as 0.0151: log likelihood = 25.1, aic = -42.3
> acf(resid(x.arma))
1.0
0.8
0.6
0.4
ACF
0.2
0.0
0.2
0 1 2 3
Lag
p}O@D MQv Q
| U |= @ Q ARMA(1,1) Q
| U x vO =t}k=@ Q=ov|oDU@ty %33 pmW
sQ=yJ pYi
=DU}==v |=ypOt
OvwQ w |rYi C= Q}eD R= QF-=Dt =Q} R 'OvDUy =DU}==v =y|QU R= |Q =}U@ Ot xDWPo = pYi
| y QO xmQw]u=ty
u}= QO "OW p}O@D =DU}= |QU x@ p =iD Q=@ l} =@ xm w =yv
"O @ R= |m} |iO=YD uORs=o p=Ft u= wvax@ "OvW=@|t
"CU= l QLDt u}ov=}t w w}UQoQwD= qtH pt=W xm O@=}|t \U@ |iO=YD
C uORs o p t= O pYi
"O W w |t xO}t=v 3 ARIMA Q=YDN= x@ xm 'CU= 2 |oJQ=Bm} =} w
1 `}tHD G =DLt |r =iD | U Q
SARIMA w x
CQ Y @ =Q u u= D w |t w OR U= |t =DU}==v =Q u xm CU= |rYi C qtH pt=W ARIMA Ov Q } i
O = |t |v=tR
" W @ = Q
| y| U R= |Q =}U@ p}rLD QO | Ovtv=wv Q=R@= |rYi ARIMA = O
| yp t "O=O u =Wv
`}=W |r=t |v=tR =
| y| U QO Q ?r]t u}= "Ovm|t Q}eD Kw w Qw]x@ Tv=} Q=w =DU}==v =
| y| U R= Q |a@ QO
w}UQoQwD= pOt '=y|QU `wv u}= |R=UpOt |=Q@ =yOQm} wQ R= |m} "OQ=O RwQ@ u=mt= R}v |t}rk= |=yOQwmQ QO "CU=
5 GARCH Q =YDN= x@ u xDi=} s}taD `wv w |t xO}t=v 4 ARCH pOt Q=YDN= x@
"O W w CU= Tv=} Q=w |= @ Q
w |t xO}t=v
"O W
1. aggregated 2. integrated 3. AutoRegressive Integreted Moving Avrage 4. AutoRegressive

Conditional Heteroskedastic 5. Generalised AutoRegressive Conditional Heteroskedastic
39
|rYi Q}e ARIMA pOt 1 4
jQ@ O}rwD |QU w p=iD 1

1
4
w
x= N w |iO=YD uORs o = pFt OW=@ |iO=YD x=wN 'Ovm P
h L =Q OvwQ Ov=wD|t f xt g |QU Q@ p =iD p=ta=
p =iD u}rw= 'xt = Xt 1 + wt |va
; } |iO=YD =
uORs o QO " CU= |]N OvwQ xm |OQ t w pFt OW=@ u}at
O = |t =DU}= xm CU=
" W @ xt = xt xt 1 = wt
r ; ;
r xt = xt xt 1 =o C O}i x w = =] = |] O l x
; ; x v ' U= U i v| y N @ N vwQ } QO m xt = a + bt + wt Ov Q p =k } i @ t QO
Q R
VR= @ arima() ` =
QO =W MA(1) =@ D "O=O u w | x C =DU v QLD u}o =} Ov Q l
@ =Q u u= D t m U= }= l t v t } i }
'O} Q R \ w u}a O Q |r =i O O} wN =t Q
yO VR= @ U D t vwQ =t |t =H C = Ok = = O
@ D p t y= @ W o = "OQ W v R t @ F Q= t @ =Q yp t
xO=iD = |r =i Q Q ARMA O
U= @ D | U w | xD@
@ =Q w | s}_v Qi u}o =} = |r =i Q =o
p t u= D t r= "O W t D Y v t @ D | U x v
Q d=0 include.mean=FALSE arima() ` =

"O=O VR= @ w w @ D R=
r xt = yt = b + wt + wt 1 u Q =v O = xt = a + bt + wt x O}v Q ARIMA O = =@
; }= @ @ ' W @ m m Z i p t @ \ DQ= QO
Pt
Q MA(1) Q O | C x xt = x0 + i=1 yi x] yt = xt
@ o = " } t UO @ wma h Qa =iD = C @=Q r T t } D R= xO U= @ " U=
O = x}@
x W |R U xt T = u Q =v C -1 Q Q wt 1 ? Q O = Q yt |r =i | = Q
W f g v } Q=w }= @ @ " U= @= @ ; } ' @ } VR= @ f g D v tR | U
"Ci=} Oy=wN V}=Ri= s}kDUt \N pwL

" O}vm xOy=Wt Q} R QO O}v=wD|t =Q Q}N= Q =
h= o =Q B C } H =} R
yt = xt r
= xt xt 1 ; ;
= a + bt + wt ; f a + b(t 1) + wt ; ; 1g
= b + wt + wt ; 1
t
X t
X
) x0 + yi = x0 + (b + wi ; wt 1) ;
i=1 i=1
= x0 + bt + wt = xt
41 =DU}==v =
| yp t O 4 pYi
"OO o Q |t x0 = a w CU= w0 = 0 x m O}W=@ xDW=O xHwD

w |t
"O W xH}D xt 1 Q
v ; |= @ w QUB |v} Ro}=H =@
xt = xt ; 1 + b + wt + wt 1 ;
= xt ; 2 + b + wt 1 + wt 2 + b + wt + wt 1
; ; ;
= xt ; 2 + 2b + wt + (1 + )wt 1 + wt 2 ; ;
= xt ; 3 + 3b + wt + (1 + )wt 1 + (1 + )wt 2 + wt ; ; ; 3
...
t
X
= x0 + bt + wt + (1 + ) wt
i=1
%R= CUDQ=@a Tv=} Q=w
Var(xt ) = w2 f1 + (1 + )2(t ; 1)g
Q@=Q@ Tv=} Q=w CQ Y w u}= QO xm O = = ;1 x u Qo O = | V R R}
W @ m }= t ' @ } t }= i= v Vv=} Q=w t Q= Okt V}=Ri= =@ xm
2
CU= w
"
sD} Q=or = =F
| yxO=O p t |= @ "O Q Q}o|t CQ Y w di \ U D w R QO p =iD u}rw= 'OW x_Lqt ,q@k xm Qw]u=ty
| U 'Q |v=tR Q
| U 14 pmW Q=O wtv xU Q} R = O O Q}o@ Q_v
| y )
m " } QO =Q xDU} QDmr= O}rwD x@ \ @ t w Q |a}@]
p t =F |r =iD Q
| U wO QO O W w |t xOy=Wt xm Qw]u=ty "OyO|t u=Wv =Q sD} Q=or |r =iD Q
| U w |r =iD
"O Ww |tv Qy=_ Qo}O OvwQ xirw-t 'xOW xDio
> www <{ "http://www.massey.ac.nz/~pscowper/ts/cbe.dat"

> CBE <{ read.table(www, he = T)
> Elec.ts <{ ts(CBE, 3], start = 1958, freq = 12)
> par(mfrow=c(3,1), mar=c(4,4,2,4))
> plot(Elec.ts)
> plot(di(Elec.ts))
> plot(di(log(Elec.ts)))
6 |k}irD pOt 2
1
4
=@ w Ov} wo d x@ Q
D t R= |k}irD pOt =Q u x =ov Q
'OO o f wt g O}iU xiwv p =iD u}t=d R= TB f xt g |QU Qo =
Q
| U " W @ O = |t wQUB p=kDv= Qorta B u QO xm CU= r
d (1 ; B )d x m s}O}O "O Ww |t xO=O u =Wv I (d)
Qo = 'CU= d x@ Q D t R= |k}irD fxt g
(1 ; B ) = wt d
=F
p t u= wvax@ "CW=O R}v q=@ x@DQt =@ =Q di() ` = w | C I (1) = C = | =Y
@ D u= D t " U= X N r L iO = O =
D uORs o " W @
u= wva Q} R `@=D u}= QO | Qo}O u=twoQ " C Q p = di(x, d=2) = di(di(x)) p

U= mP @ k } w =iD Q=@ wO |= @Q
6. Integrated model
14000
Elec.ts
8000
2000
1960 1965 1970 1975 1980 1985 1990
1000 Time
diff(Elec.ts)
0
1500
1960 1965 1970 1975 1980 1985 1990
Time
0.00 0.10 0.20
diff(log(Elec.ts))
0.15
1960 1965 1970 1975 1980 1985 1990
Time
|tD} Q=or |r=iD Q

| U w |r=iD | U 'Q |v=tR | UQ V}=tv %14 pmW
u}= "CU= OL=w Q@=Q@ ZQiV}B CQwYx@ u Q= Okt "Ovm XNWt =Q p =iD Q}N-=D Ov=wD|t xm CU= OwHwt lag
xirw-t di(x, lag=12) =F Q
p t |= @ "OO o Q |atH pOt QO |rYi C= Q}}eD xirw-t h L P ?@U Ov=wD|t u=twoQ
Cw =iD CU= umtt =Hv}= QO "O}=tv|t hPL xv=y=t |=y|QU QO =Q |atH pOt |rYi C=Q}}eD QF= w |]N OvwQ
u O)m TBU 'O}vm xHwD wt = vt vt ; ; 3 =F x@ s=y@= `iQ
p t Q O = xDW=O s=y@=
|= @ " W @ lag d w = w
u t oQ wO u}@
" CU= Q} R CQ Y @w x di() ` = @ D QO
> w <{ di(v, lag=2, d=1)
=yp=Ft w h} QaD 3
1
4
Ov} Qi x@ p}O@D xt
f g |QU p =iD u}t=d Qo = 'CU= ARIMA(p,d,q) Ov Q } i |=Q=O f xt | = Q
g v tR | U
= x Qo = uwvm = "CU= p (
| H @
B )yt = q (B )wt x =ov O = yt = (1 ; B )d xt Q
' W @ o = "OO oQ ARMA(p,q)
xm OwW|t xH}Dv 'O Q} Q xt Q}eD yt
o Q= k t

p(B )(1 B )dxt = q (B )wt
;
ARIMA = O
| yp t R= p t =F OvJ x@ uwvm = "OvDUy q w p x@DQt R= ?}DQD x@ |}=y|=xrtH OvJ q w p
u QO xm
" O}vm xHwD
43 =DU}==v =
| yp t O 4 pYi
wQUB p=kDv= Qorta w x

CQ Y @ u wvm = "CU= pOt QDt=Q=B u QO x = xt 1 + wt + wt
xm t ; 1
; Ov} Qi
"O Ww |t xDWwv
xt xt 1 = wt + wt
; ; ;1 ) (1 ; B )xt = (1 + B )wt
l QLDt u}ov=}t |k}irD pOt u x@ w ' CU= ARIMA(0,1,1) w x
CQ Y @ f xt Q g | U 'xU}=kt =@ u}=Q@=v@
ARIMA(0,d,q)IMA(d,q) |r m Cr=L w |t
QO "O W xO=O u =Wv IMA(1,1) w x xm Ov} wo
CQ Y @
"CU=
Qorta CQ Y @ uw x wvm = "CU= pOt QDt=Q=B u QO xm t x = xt 1 + xt

; ; 1 ; xt 2 + wt Ov Q
; } i
w |t xDWwv wQUB p=kDv=

"O W
xt xt; ; 1; xt 1 + xt 2 = wt
; ; ) (1 ; B )(1 ; B )xt = wt
|rm Cr=L QO "O W w |t xO=O u =Wv ARI(1,1) w x xm Ov} wo |k}irD w}UQoQwD=
CQ Y @ O
p t u x@ w
" CU= ARI(p,d)ARIMA(p,d,0)
VR=Q@ w |R=Ux}@W 4
1
4
u}}aD c(p,d,q) = O x@ Q x O = |
@ p t Q R
D t arima() ` =
m @ } t VR= @ QO @ D =@ =yxO=O Q@ ARIMA(p,d,q) Ov Q } i
|R U = x}@ xt = 0:5xt 1 + xt 1 + wt + 0:3wt 1 x w Q

W ; ; = ; @ \ @ t | yxO=O Q} R |=yO)m =F
QO p t |= @ "OO o Q Q |t
"O Ww |t xO=O VR= @ Q =yv Q@ ARIMA(1,1,1) p t O TBU w x WO
> set.seed(1)
> x <{ w <{ rnorm(1000)
> for (i in 3:1000) xi] <{ 0.5 xi - 1] + xi - 1] - 0.5
+ xi - 2] + wi] + 0.3 wi - 1]
> arima(x, order = c(1, 1, 1))
Call:
Coecients:
ar1 ma1
0.4235 0.3308
s.e. 0.0433 0.0450
"OQ=O s v = arima.sim() ` = @ D u}= "OyO|t s=Hv= = O)m u=ty Q=m xm
=Q q @ w
OQ=O O Hw |= v x =N@=Dm `@=D R QO xD@r=
"CU= Q} R w x O
CQ Y @ x W |R U = x}@W sDU}U uwvm =
> x <{ arima.sim(model = list(order = c(1, 1, 1), ar = 0.5,

+ ma = 0.3), n = 1000)
> arima(x, order = c(1, 1, 1))
Call:
Coecients:
ar1 ma1
0.5567 0.2502
s.e. 0.0372 0.0437
|rYi ARIMA |=ypOt 2 4
h} QaD 1
2
4
|atH pOt |rYi C=Q}}eD QF= =D Ovm|t xO=iDU= (s) pYi xR=Ov= x@ |Q}N-=D =@ p =iD R= ARIMA |rYi pOt
u}ov=}t xrtH Qov=}@ p =iD s Q} =
N-D w Ovm|t h L =Q P OvwQ xirw-t p =iD Qorta OL=w Q}N-=D =@ Q
"OO o h L P
O
p t " W @ O = |t s Q} = =@
N-D l QLDt u}ov=}t w w}UQoQwD= C qtH pt=W ARIMA |rY O C QLD i p t " U= l t
CWwv Q} R w x QUB p=kDv= Qorta ?UL Q@

CQ Y @ w u= Dw |t =Q ARIMA(p d q)(P D Q)s |rY i
P (B s)
p(B )(1 ; B s)D (1 ; B )d xt = Q(B s) q (B )wt
|twta Cr=L QO " OvW=@|t q Qp P
w ' ' x@DQt R= | y= xrtH OvJ ?}DQD x@ q Q
p P
w ' ' u QO xm
|= xrtH OvJ C qtH xYNWt xrO=at | y = xW} Q jr]t QOk w D=d=0Q o= =t= "CU= =DU}==v pOt u}=
|rYi pOt p=Ft OvJ x@ uwvm = "O @ w Oy=wN =DU}= x=ov 'OvW=@ OL=w R= QDoQR@ |oty Q}N= xrO=at AJ Q
h ]
"O}vm xHwD ARIMA

xt = x12 + wt CQ Y @w x xm ARIMA(0 0 0)(1 0 0)12 12 |rY ' i ?w D=v =@ AR =
xO U p t O
=
p U x t |wQ = xDWPo p=U x=t xm u Q@ \w QWt CU= ?U=vt xv=y=t =
| yxO=O |= @ Q |rOt u}vJ "CU=
O =
" W @ j ; 1=12
j >1Q o= CU= =DU}= pOt u}= "OW=@ xDW=Po QF= |Q H =
w |}=H@=H =@ x = xt 1 + xt 12 xt 13 + wt w x w | = Ov O
"O=O \U@ t ; ; ; ; CQ Y @ u= D t =Q q @ @ p t
O | C x 1 (B 12 )(1 B )xt = wt = (1 B 12 )(1 B )xt = wt x]

' } t UO @ ; Q} wD = } w ; ; @=Q | oQ m i
w | p = ARIMA(0 1 0)(1 0 0)12 Ov Q |rY ARIMA |r x] = xU =k = x

"O W t Y L } i ' i m @=Q @ } t @ m
?r] u x x w = C w w | R}
t }= @ xt = xt 12 + wt w x O u x O}v x w
H D @ " W v u= D t v r r ; CQ Y @ =Q p t }= m m H D
C =DU = O u
' U= xD P =
}= v p t = |va = =t
}= "OQ=OQ}}e x |oDU t =
W o p U QO x t Q}}e x } u tR u y QO D @ @ u tR QO D m
O = | B = 1 xW " W @ x C (1 B ) xrt p = A
t Q xrt Ov Q
} Q |=Q=O m U= ; H t W J h ] |= H J = }R
45 =DU}==v | yp t = O 4 pYi
u}= "CU= t x = (1 B 4)wt = wt wt

; ; 4
; w x pYi Q=yJ
CQ Y @ l QLDt u}ov=}t pOt l}
'OvW=@ |iO=YD OvwQ |=Q=O =yxO=O Qo = "CU= ?U=vt OvwQ O

uw @ | yxO=O |= @ = Q \ki w CU= =DU}= pOt
w x
CQ Y @ u= D w |t w 'CU= t x = xt 1 + wt wt
; ; ; 4 xm O}=}|t \U@ pw= x@DQt p =iD CQwYx@ pOt
?w D QO =v p =iD 'OW=@ |iO=YD OvwQ |=Q=O |rYi CqtH Qo = "CWwv R}v ARIMA(0 1 0)(0 0 1)4
w x
CQ Y @ =Q u w ' yO O |t CUOx@ =Q xt = xt 4 + wt wt
; ; ; 4 x]@=Q w OO o Q |t p=ta= R}v |rYi
" CWwv u= Dw |t R}v ARIMA(0 0 0)(0 1 1)4
CW=O l} Q}N-=D =@ p =iD u= D w |t u}=Q@=v@ 'Ovm|t h L =Q P |]N OvwQ s Q}
' N-D @ = = p =iD xm CW=O xHwD O}=@
l QLDt u}ov=}t CqtH x=ov 'OQ=O OwHw |]N OvwQ xm |r=L QO OOQo p=ta= p =iD QO l} Q}N-=D Qo = #Q}N =}
"OO oQ |t |iQat O}iU xiwv w x
CQ Y @
p t O l} CQwYx@ O}iU xiwv w |rYi C=Q}}eD w |]N OvwQ u QO xm 4 ?w D @ =v = |v=tR |QU x@ 'p t u= =F wvax@
" O}vm xHwD OQ=O O Hw w |atH
xt = a + bt + s
t] + wt
" O}vm xHwD 4 Q} = = p =iD u}rw= x@ uwvm =
N-D @ s
4] = s
t ; 4] u}=Q@=v@ "OyO|t u=Wv =Q 4Q t @ O =t}k=@
x v t] xm
(1 ; B 4 )xt = xt ; xt 4 ;
= a + bt ; (at + b(t ; 4)) + s

t] ; s
t ; 4] + wt ; wt ;4
= 4b + wt ; wt 4 ;
u}rw= xm O}vm Q
Z i u wvm = "O wtv u=}@ 4b C = @ F xrtH =@ ARIMA(0,0,0)(0,1,1) w x CQ Y @ =Q u u= D w | t xm
=ov
x Q
'OO o p =t 4 Q} = = p =i
a= p@ l Q} =
N-D @ D R= k } N-D @ = p =iD x@DQt
(1 ; B 4 )(1 ; B )xt = (1 ; B 4)(b + s

t] ; s
t 1] + wt ; wt 1) ; ;
= wt ; wt 1 ; wt 4 + wt 5 ; ; ;
" CU= C@=F Q=Okt Ok=i xm CWwv ARIMA(0 1 1)(0 1 1)4 w x

CQ Y @ =Q u u= D w |t xm
VR=Q@ x} wQ 2
2
4
u}=Q@=v@ "OW=@ CqtH R= |@}mQD w OOaDt |=yQDt=Q=B pt=W Ov=wD|t xwkr=@ CQwYx@ '|rYi ARIMA |=ypOt
R= QD?U=vt pOt ? =NDv= Q w
|= @ w O W u =LDt= =yxO=O Q@ Q
VR= @ |= @ p t R=Q O |a}Uw xOw OLt xm CU= xDU}=W
xiwv CQwYx@ xOv=t}k=@ Q=ov|oDU@ty |=Q=O ,=k]vt 'xOW xDi=} VR=Q@ ?U=vt pOt "OOQo xO=iDU= AIC x]@=
" CU= O}iU
xm OQ=O O Hw w seasonal u= wva Q} R |v=twoQ `@=D u}= QO "O W w |t xO =iDU= VR= @ |= @Q Q arima() ` = @ D R= ' R QO
x W O xDiQo Q_v QO xDU} QDmr= O}rwD =

| yxO=O sD} Q=or | U Q Q} R p=Ft QO "OO o Q |t GQO u QO |rYi | y = xirw-t
=Q |QDy@ VR= @Q ARI O

p t " Ovm|t P
h L =Q |]N OvwQ xm CU= d=1 w
OQ t wO Qy QO p =iD QDt=Q=B "CU=
"OQ=O | QDmJwm AIC Q
= } R ' yOO |t u=Wv

> AIC (arima(log(Elec.ts), order = c(1,1,0),
+ seas = list(order = c(1,0,0), 12)))
1] -1764.741
> AIC (arima(log(Elec.ts), order = c(0,1,1),
+ seas = list(order = c(0,0,1), 12)))
1] -1361.586
`@=D Qt= u}= QO CrwyU Q
|= @ "O Ww |t xO =iDU= =]N w |aU R= QD?U=vt pOt uDi=} w =ypOt xvt=O pQDvm Q
|= @
VwQ |Dkw OQm} wQ u}= "OyO CUOx@ =Q QD?U=vt pOt AIC x] = =

@ T U= Q@ =D CWwv u= Dw |t =Q |mJwm
O}vm xHwD Q} R
" | y )
m= O x@ uwvm = "CU= Q wDU= sD} Qwor=
| DQ= w OyO|t ?= Hw QDy@ OW=@ CSS
> get.best.arima <{ function(x.ts, maxord = c(1,1,1,1,1,1))
+f
+ best.aic <{ 1e8
+ n <{ length(x.ts)
+ for (p in 0:maxord1]) for(d in 0:maxord2]) for(q in 0:maxord3])
+ for (P in 0:maxord4]) for(D in 0:maxord5]) for(Q in 0:maxord6])
+f
+ t <{ arima(x.ts, order = c(p,d,q),
+ seas = list(order = c(P,D,Q),
+ frequency(x.ts)), method = "CSS")
+ t.aic <{ -2 t$loglik + (log(n) + 1) length(t$coef)
+ if (t.aic < best.aic)
+f
+ best.aic <{ t.aic
+ best.t <{ t
+ best.model <{ c(p,d,q,P,D,Q)
+g
+g
+ list(best.aic, best.t, best.model)
+g
> best.arima.elec <{ get.best.arima( log(Elec.ts),
+ maxord = c(2,2,2,2,2,2))
> best.t.elec <- best.arima.elec2]]
> acf( resid(best.t.elec) )
> best.arima.elec 3]]
1] 0 1 1 2 0 2
47 =DU}==v | yp t = O 4 pYi
> ts.plot( cbind( window(Elec.ts,start = 1981),

+ exp(predict(best.t.elec,12)$pred) ), lty = 1:2)
QDq=@ ?D=Qt uwtR "CU= ARIMA(0 1 1)(2 0 2)12 swO x@DQt R= QDC@U=vt pOt =
'q @ | y )
m |= H= = O Q =@
pmW "OvDUy 34 pmW O}iU xiwv ,=@} QkD xOW xDio pOt = O =t}k=@ =Q} R 'OUQ|tv Q_v x@
| yx v Q
|Qw p t O
O
" yO |t u=Wv =Q x W O |v}@V}B Q}O=kt 24
14000
12000
10000
8000
1982 1984 1986 1988 1990 1992
Time
Q O}rwD xv=y=t
j @ Q
| U |v}@V}B V}=tv %24 pmW
ARCH |=ypOt 3 4
SP500 | y| U = Q 1
3
4
"OQ=O O Hww R QO MASS x =N =D

v @ m QO McGraw-Hill | =Bt v m x@ w Q
\ @ t SP500 =
| yxO=O
> library(MASS)
> data(SP500)
> par(mfrow=c(2,1), mar=c(4,4,2,4)) > plot(SP500, type = 'l')
> acf(SP500, main="")
=@ "OyO|t u=Wv =Q =DU}= Ov} Qi pw= x =ov O
QO " yO |t u=Wv O
=Q x W xDio C a = q]= |v=tR Q
| U 44 pmW
Q=Okt u} QDoQR@ 'QN ErF QO u Q= Okt w u} QDmJwm Q
| U \Uw ErF QO Tv=} Q=w xm OwW|t swrat QDW}@ CkO
Series resid(best.fit.elec)
1.0
0.8
0.6
ACF
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0
Lag
Q O}rwD xv=y=t
j @ Q
| U | yx v = O =t}k=@ V}=tv %34 pmW
w OW=@v C@=F u=tR x@ C@Uv Tv=} Q=w Qo = "O Ww |tv s=Hv= |t_vt Q}Ut QO V}=Ri= Qo}O =@ x@
CQ a "OQ=O =Q
x@ x=ov O@=}|t V}=Ri= OvwQ =@ Tv=} Q=w xm j @ Q O}rwD w |}=t}B=wy =

| yxO=O Ovv=t 'OQ}o s=Hv= s_vt C= Q}}eD
xO=O pFt Q
OO o u tR = x@ xDU@=w w OW=@ |W}=Ri= Ow} QB |=Q=O | U Q heteroskedastic
Tv=} Q=w Qo = "Ov} wo u
Q}eDt Q
| U Q =ov|oDU@ty xm O}W=@ xDW=O xHwD "Ov} wo |]QW heteroskedastic Q x =o SP500| U @ x v
Cw =iDt u=tR QO xm u}= p}rO x@ CU= =DU}==v |QU =t= 'OQ=Ov O}iU xw = |va
i v @ |Q=O =i SP500 = Ct}
t Cw D y k
`@ Qt Q=ov|oDU@ty x=ov 'OyO|t u=Wv =Q O}iU xiwv Q=ov|oDU@ty ?? pm W j@=]t Qo = "CU= Q}eDt Tv=} Q=w
u}ov=}t xm u \ WQ x@ OvDUy Tv=} Q=w pO =at C =a@ Qt u}= "OyO|t u=Wv =
=Q u tR x@ C@Uv C= Q}}eD Q}O=kt
0.048 Q Q 1999 Q@ = 31 = 1900 x w 2 SP500
@= @ t UO D } v=S R= xO=O u}ov=}t "OO o Q s}_vD QiY CQ Y @ w x =yv
w | xD Q Q_
"O W t i o1469 = 360 T O u Q =v C
v QO D R= } v= }= @ @ " U= |mJwm Q=Okt Tv=} Q=w =@ xU}=kt QO xm 'CU=
C Q w x SP500 O
" U= } R CQ Y @ x W K}LYD u}ov=}t =@ `@ Qt Q}O=kt Q=ov|oDU@ty
library(MASS)
data(SP500)
acf((SP500 - mean(SP500))^2)
" yOO |t u=Wv O==
=Q |Q= } B v w |]QW heteroskedastic w
K w x@ 54 pmW
49 =DU}==v = O
| yp t 4 pYi
2 4
SP500
2
6
0 500 1000 1500 2000 2500
Index
0.8
ACF
0.4
0.0
0 5 10 15 20 25 30 35
Lag
SP500 =
| yxO=O | U Q V}=tv %44 pmW
Series (SP500 mean(SP500))^2

1.0
0.8
0.6
ACF
0.4
0.2
0.0
0 5 10 15 20 25 30 35
Lag
O
x W K}LYD u}ov=}t =@ Q}O=kt `@ Qt V}=tv %54 pmW
ARCH p t O h} QaD %|Q=O}=B=v uOQm pOt 2

3
4
x@ uO}UQ Q Ovm u}@D

|= @ " =Q Tv=} Q=w QO |]QW C= Q}}eD xm O=O CUOx@ |rOt O}=@ =D
'Q iQ u}= Q
uO m p t |= @ O Q
xO=O u =Wv ARCH(1) = @ xm CU= heteroskedastic | ] W Q w}UQoQwD= x@DQt u}rw=
f g |QU '?r]t u}=
Qo = 'O W w |t
q
t = wt 0 + 1 2t ; 1
OvDUy pOt
" | y = QDt=Q=B 1 0 ' w CU= OL=w Tv=} Q=w w QiY u}ov=}t =@ O}iU xiwv f t g q=@ x]@=Q w O =
QO " W @
" }O Q}o@ Tv=} Q=w Q}N= x]@=Q R= volality =}

u @ |= @Q
Var(t ) = E (2t )
= E (2t )E ( 0 + 1 2t 1 ) ;
= E ( 0 + 1 2t 1 ) ;
= 0 + 1 Var(t 1 ) ;
AR(1) Ov Q =
} i @ =Q Q}N= xrO=at Qo = 'CU= QiY u}ov=}t |=Q=O f g w w
OL=w Tv=} Q=w |=Q=O f t g xm u}= x@ Q_v
O}vm xU}=kt
xt = 0 + 1 xt 1 + wt ;
O v m| t Q= D iQ
" AR(1) p F t C UQO ARCH(1) O v Q T = x O } v | x _ q
} i v } Q=w m m t L t "O Ww | t p Y= L
# CU= ?U=vt ARCH O Q x Ov | XNW = =] ` Q
p t VR= @ m m t t y N @ t | y = |oDU@tyOwN
GARCH | yp t = O x@ \w@ Qt |=y\U@ 3

3
4
w x
CQ Y @ ARCH(p) Ov Q } i " @ } O = \U@ q=@ |=yQ}N-=D =@ p x@DQt Ov} Qi l} x@ Ov=wD|t ARCH pOt x@DQt u}rw=
"CU= Q} R
v
u p
u X
t = wt 0 + p2t;i
t
i=1
" CU= OL=w Tv=} Q=w w QiY u}ov=}t =@ O}iU xiwv fwt g u QO xm
ARCH(p) x m ' CU= GARCH(q,p) CQ Y @w x ARCH p t R= O |r=t |=yxO=O OQ@ Q=m =@ QD`}Uw |=y\U@
Qo = 'OW=@|t GARCH(q,p) O
p t |=Q=O f t g |QU "CU= GARCH(0,p) X N = Cr=L
p
t = wt ht
u QO xm
p
X q
X
ht = 0 + i2t i +
;
j ht ; j
i=1 j =1
OvDUy pOt
" | y= QDt=Q=B j (j = 0 1 : : : q) i (i = 0 1 : : : p)w w CU=
51 =DU}==v =
| yp tO 4 pYi
xDi=} VR=Q@ GARCH p t w |R UO = x}@W 4

3
4
ht = 0 + t 1 + t a = wt ht
p
; ; 1 xm t w x
CQ Y @ GARCH(1,1) O =@ |}=yxO=O Q} R
p t = O
| y )
m QO
CU= 64 pmW

" w x =ov|oDU@ty w OW=@
CQ Y @ Q 1 + 1 < 1 O = O = Q
} @ |Q= } B |= @ "O W w |t = x}@W CU=
|R U
> set.seed(1)
> alpha0 <{ 0.1
> alpha1 <{ 0.4
> beta1 <{ 0.2
> w <{ rnorm(10000)
> a <{ rep(0, 10000)
> h <{ rep(0, 10000)
> for (i in 2:10000) f
+ hi] <{ alpha0 + alpha1 (ai - 1]^2) + beta1 hi -1]
+ ai] <{ wi] sqrt(hi])
+g
> par(mfrow=c(2,1), mar=c(4,4,2,4))
> acf(a)
> acf(a^2, main="")
Series a
0.8
ACF
0.4
0.0
0 10 20 30 40
Lag
0.8
ACF
0.4
0.0
0 10 20 30 40
Lag
=yv `@ Qt w Q}O=kt Q=ov|oDU@ty V}=tv %64 pmW
Qy=_ |oDU@ty Q}O=kt `@ Qt QO xm OQ=O xDU@ty=v Q}O=kt xm OyO|t u=Wv =Q GARCH p t X= NO w a | UQ

"OO oQ |t
garch() ` = @ D R= xO =iDU= =@ xm O@=}|t Q O

VR= @ x W |R U = x}@W =
| yxO=O |wQ Q@ GARCH O Q
p t ' } R p t QO=F
|@=} R=@ xQ=@ wO pOt |=yQDt=Q=B xm OOQo|t x_Lqt "OwW|t Ci=} tseries R
|Q= i=s v Q xDU@ QO xm OwW|t s=Hv=
= w xm CU= GARCH(1,1) Q@=Q@ Q}N= `@=D
,a v Z i Q V}B "OQ}o|t Q=Qk %95 u=v}t]= =
xR @ xOw OLt QO w O Ww |t
w |t
"O W =iDU=
xO order=c(p,q) = w
u t oQ R= QDq=@ | y = x@DQ |= @ Q =t= 'CU= woNU=B
> set.seed(1)
> alpha0 <{ 0.1
> alpha1 <{ 0.4
> beta1 <{ 0.2
> w <{ rnorm(10000)
> a <{ rep(0, 10000)
> h <{ rep(0, 10000)
> for (i in 2:10000) f
+ hi] <{ alpha0 + alpha1 (ai - 1]^2) + beta1 hi -1]
+ ai] <{ wi] sqrt(hi])
+g
> library(tseries)
> library(quadprog)
> library(zoo)
> a.garch <{ garch(a, grad = "numerical", trace = FALSE)
> connt(a.garch)
Q |t pY=L Q} R G}=Dv
"OO o =
'q @ | y )= O
m |= H= Q =@
2.5 % 97.5 %
a0 0.0882393 0.1092903
a1 0.3307897 0.4023932
b1 0.1928344 0.2954660
w x
CQ Y @ =Q |O a C U O =@ =Lt pL=Qt w OyO|t CUOx@ =Q C U =@ =Lt xYqN trace=F = w =F
u t oQ p t QO
O
" yO |t CUOx@ Q wDU=
=Q | DQ= O
|O a VwQ grad="numerical" = w
u t oQ " yO O |tv u=Wv |r}iD
SP500 | U Q Q@ VR=Q@ 5
3
4
Q} R x]@=Q R= fwt g |va} GARCH O

p t x v O =t}k=@ Q
| U " @ }O = |t VR= @ Q SP500 | U Q Q@ GARCH O
p t
w |t x@U=Lt
"O W
w^t = pt
h^ t
QiY u}ov=}t =@ O}iU xiwv w x O}=@ xH}Dv x=ov 'OW=@ ?U=vt =yxOv=t}k=@ |QU
CQ Y @ |= @ Q GARCH O Qo =
p t
GARCH(1,1) C = r L QO "O W w |t Qy=_ OL=w Tv=} Q=w w
h^ t = ^0 + ^12t 1 + ^h^ t
; ; 1
53 =DU}==v | yp t= O 4 pYi
"CU= t = 2 : : : n x h^ 1 = 0 |=R= @ u QO xm
=@ Q= Okt u u}=Q@=v@ "CU}v OwHwt xOv=t}k=@ Q
| U Q= Okt u}rw= "OQ} | =H garch ` = \ w
o t s v= @ D U D C U =@ =Lt
O |t CUOx@ =yxOv=t}k=@
" } C =a@ Qt Q
| U w x v O =t}k=@ Q
| U Q =o |oDU@t
v w | P -1]
y w O W t h L R= xO =iDU=
s}rk= |QU QO |Q=O}=B=v 6

3
4
C =k}kLD OL=w x@ w Q
\ @ t 1850-2007 =
| yxO=O " CU= xOW s=Hv= |D=ar=]t s}rk= O = = x@ `H=Q ,=Q}N=
QO |Q= } B v
O s}UQD
x W 74 pmW O
QO w x W x v= N O w Q_v w Q
OQ t | U " CU= xDiQo Q=Qk pta l qt East Anglia =oW x v=O
CUOx@ 'CU= xOt 224 CtUk |=yO)m QO xm get.best.arima `@=D \UwD ?U =v ARIMA O C
t p t " U=
" CU= xOW p}O@D |vwDU x@ | Q]U Cr=L R= t() ` =@ D R= xO =iDU= =@ pOt | y = QDt=Q=B Q
|= @ u =v}t]= = O |t
xR @ " }
> stemp <{ scan("http://www.massey.ac.nz/~pscowper/ts/stemp.dat")

> stemp.ts <{ ts(stemp, start = 1850, freq = 12)
> plot(stemp.ts)
> stemp.best <{ get.best.arima(stemp.ts, maxord = rep(2,6))
> stemp.best3]]
1] 1 1 2 2 0 1
0.5
0.0
stemp.ts
0.5
1.0
1850 1900 1950 2000
Time
s}rk= O==
|Q= } B v | U Q V}=tv %74 pmW
"OO oQ |t Q O
OQw @ p t | y = QDt=Q=B w
j i | y )
m = O xt=O= QO
> stemp.arima <{ arima(stemp.ts, order = c(1,1,2),

+ seas = list(order = c(2,0,1), 12))
> t( connt(stemp.arima) )
ar1 ma1 ma2 sar1 sar2 sma1
2.5 % 0.8317389 -1.447400 0.3256699 0.8576802 -0.02501886 -0.9690530
97.5 % 0.9127947 -1.312553 0.4530474 1.0041394 0.07413434 -0.8507036
Q
VR= @ ,=O OHt Q} R = O =@
| y )
m O P
p t = r 'OQ= v O QiY =@ |Q=O |vat Cw =iD sar2 |va AR |rY
} i xirw-t u}twO
O = |t
" @ }
> stemp.arima <{ arima(stemp.ts, order = c(1,1,2),

+ seas = list(order = c(1,0,1), 12))
> t( connt(stemp.arima) )
ar1 ma1 ma2 sar1 sma1
2.5 % 0.8304007 -1.445057 0.3242728 0.9243495 -0.9694897
97.5 % 0.9108115 -1.311246 0.4508999 0.9956639 -0.8679223
=vt "CU= xOW s}UQD 84 pmW QO
, ARIMA O
p t | yx v= O =t}k=@ Q=ov|oDU@ty 'VR=Q@ |} wmv pQDvm |=Q@
w |t x_Lqt Q}N= pmW
"O W QO R}v =yxOv=t}k=@ `@ Qt Q=ov|oDU@ty |UQQ@ Q
|= @
> stemp.res <- resid(stemp.arima)

> par(mfrow=c(2,1), mar=c(4,4,2,4))
> acf(stemp.res)
> acf(stemp.res^2, main="")
O=|" @ } Q = O =t} = Q Q GARCH
t VR= @ yx v k @ | U @ O u}=Q@=v@
p t w
'OQ=O O Hw |Q= } B v O = = xm CU= uWwQ ,qt=m
|R=Ux}@W w |v}@V}B QO GARCH 7

3
4
O
" yO |t u=Wv =yxOv=t}k=@ QO |v=tR Q
| U p}rLD w
QO =Q O N QF= GARCH p t O OW x_Lqt ,q@k xm Qw]u=ty
w xt}@ '|r=t |v=tR =
| y| U CQ =ar=]t QO p t O u}= |rY= OQ@ Q=m u}=Q@=v@ O
"OQ= v \ =kv |v}@V}B QO | Q}F-=D =Pr
" CU= |R U = x}@W Ov} Qi |= @Q s}rk=
55 =DU}==v = O
| yp t 4 pYi
Series stemp.res
0.8
ACF
0.4
0.0
0.0 0.5 1.0 1.5 2.0 2.5
Lag
0.8
ACF
0.4
0.0
0.0 0.5 1.0 1.5 2.0 2.5
Lag
u C =a@ Qt w =yxOv=t}k=@ Q=ov|oDU@ty V}=tv %84 pmW

`H=Qt
1] Cowpertwait, Paul S.P., Metcalf Andrew V. (2009) Introductory Time

Series with R, Springer, 254p.
2] Ihaka Ross, (2005) Time Series Analysis, University of Auckland, 105p.
3] Shumway Robert H., Stoer David S., (2011) Time Series Analysis and
Its Applications (With R Examples), Third edition, Spriger, 596p.

R in Time Series

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

R in Time Series

Uploaded by

Copyright:

Available Formats

CNwt CQmi =

QwBU=@a O}yW jQ@ w ? CavY x=oWv=O

100 Q=DWwv CU=Q} w

1 v tR |=y|QU pw= pYi

8 . . . . . . . . . . . . . . . . . . . . |oDU@t T = w h Qay w v } Q=w m } D 231

17 . . . . . . . . . . . . . . . . . . Q=ov|oDU@ty w swO x@DQt X= N w 412

19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .| iO=YD uORs o = 22

20 . . . . . . . . . . . . . . . . . . . . . . . . . QU =kD Qort w B p v= a 322

21 . . . . . . . . . . . . . . . . . . .......... = x}@ |R U W 622

25 . . . . . . . . . . . . . . . . . . . . AR(1) O x@ Q w p t swO D t X= N 332

26 . . . . . . . . . . . . . . . . . . . . . AR(1) Ov Q =o |oDU@t } i Q v y 432

27 . . . . . . . . . . . . . . . . . . .......... = x}@ |R U W 632

28 . . . . . . . . . . . . ... xDi=} VR= @ Q AR O

30 =DU }= |=ypOt swU pYi

37 . . . . . . . . . . . . . . . . . . . . . . . . . Q = x}@ VR= @ w |R U W 143

39 =DU}==v |=ypOt sQ=yJ pYi

43 . . . . . . . . . . . . . . . . . . . . . . . . . Q = x}@ VR= @ w |R U W 414

44 . . . . . . . . . . . . . . . . . . . . . . . . . . . . |rY ARIMA = i | yp tO 24

45 . . . . . . . . . . . . . . . . ............ Q x VR= @ } wQ 224

52 . . . . . . . . . . . . . . . . . . . . . . . SP500 Q Q Q | U @ VR= @ 534

54 . . . . . . . . . . . . . . . . . = x}@ |v} V} GARCH

x@ EL=@t u}= w O Ww |t s}OkD pYi OvJ QO u K W Q xm OQ=O |v=tR =

" CU= xO wtv ?Um =Q R R

u < =kDQ= w K qY= Q

1391 R}}=B  u= QyD

x@ Ov=wD|t O =Wt C@F pta "Ovvm|t Q}}eD u=tR

|rYi C=Q}}eD w OvwQ '|v=tR |QU s}UQD 1

> AP <{ AirPassengers

1950 1952 1954 1956 1958 1960

u} Qi=Ut xv=y=t |v=tR | U Q V}=tv %11 pmW

" O}vm xHwD Q} R p=Ft x@ "OyO CUOx@ =Q |v=tR Q

> x <{ AirPassengers

> x <{ AirPassengers

> x <{ AirPassengers

V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11 V12

" CU= Q} R KQW x@ w

> x <{ matrix(scan("F:/R_les/data/ghar.txt"), ncol=12, byrow=T)

> x <{ read.table("F:/R_les/data/ghar.txt")

O}rwD p}=i 1990 =

choc beer elec

|=Q=O w OvQ=Ov M} Q=D Qo}O CQ a =@ x@ "OvDUy x=t w p U = Ok=i q=@ =

"O}vm xHwD Q} R = O x@ Q=m u}=

cbind(Elec.ts, Beer.ts, Choc.ts)

1960 1965 1970 1975 1980 1985 1990

|v=tR | UQ OvJ V}=tv %21 pmW

1. trend 2. seasonal variation

|rY i C=Q}}eD w OvwQ 3 1 1 1

1950 1952 1954 1956 1958 1960

=yv boxplot w x W O `}tHD xvq=U Q

`}tHD `@=D \UwD xvq=U \UwDt "OO o Q |t x_Lqt u QO freq =v

> www <{ "http://www.massey.ac.nz/~pscowper/ts/Maine.dat"

> Maine.month.ts <{ ts(unemploy, start = c(1996, 1), freq = 12)

1996 1998 2000 2002 2004 2006

1996 1998 2000 2002 2004

=yv \UwDt w |Q =m}@ Q

|v=tR |=y|QU x} RHD 2 1

=ypOt u}=Q@=v@ w |t xOy=Wt |rYi

Q |t p}O@D |atH pOt x@ |@ Q pOt 'Q}N= x]@=Q

yt = ln(xt ) = ln(mt ) + ln(st ) + ln(zt ) 0 0 0

Decomposition of multiplicative time series

1960 1965 1970 1975 1980 1985 1990

|oDU@ty u}vJ O}=@ =yv p}rLD w x} RHD Q

QwBU=@a O}yW jQ@ w ? CavY x=oWv=O

100 Q=DWwv CU=Q} w

8 . . . . . . . . . . . . . . . . . . . . |oDU@t T = w h Qay w v } Q=w m } D 231

17 . . . . . . . . . . . . . . . . . . Q=ov|oDU@ty w swO x@DQt X= N w 412

19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .| iO=YD uORs o = 22

20 . . . . . . . . . . . . . . . . . . . . . . . . . QU =kD Qort w B p v= a 322

21 . . . . . . . . . . . . . . . . . . .......... = x}@ |R U W 622

25 . . . . . . . . . . . . . . . . . . . . AR(1) O x@ Q w p t swO D t X= N 332

26 . . . . . . . . . . . . . . . . . . . . . AR(1) Ov Q =o |oDU@t } i Q v y 432

27 . . . . . . . . . . . . . . . . . . .......... = x}@ |R U W 632

37 . . . . . . . . . . . . . . . . . . . . . . . . . Q = x}@ VR= @ w |R U W 143

43 . . . . . . . . . . . . . . . . . . . . . . . . . Q = x}@ VR= @ w |R U W 414

44 . . . . . . . . . . . . . . . . . . . . . . . . . . . . |rY ARIMA = i | yp tO 24

45 . . . . . . . . . . . . . . . . ............ Q x VR= @ } wQ 224

52 . . . . . . . . . . . . . . . . . . . . . . . SP500 Q Q Q | U @ VR= @ 534

1391 R}}=B u= QyD

u} Qi=Ut xv=y=t |v=tR | U Q V}=tv %11 pmW

> x <{ matrix(scan("F:/R_les/data/ghar.txt"), ncol=12, byrow=T)

> x <{ read.table("F:/R_les/data/ghar.txt")

|v=tR | UQ OvJ V}=tv %21 pmW

|rY i C=Q}}eD w OvwQ 3111

|rYi C= Q}}eD w OvwQ V}=tv %61 pmW

(x y) = E (x x)(y y )] ; ;

(x y) = E (x x)(y y )] = (xy)

2 (t) = E (xt )2] ;

|oDU@t yOwN x@U=Lt 1331

(xt ; x) (xt+1 ; x)

(xt ; x) (xt+k ; x)

"O Ww |t xO =iDU= =ypw] QwLt k ?U Q =

u Q =ov|oDU@ty w |v=tR | U Q V}=tv %81 pmW

> acf(waveht, type = c("covariance"))$acf2]

"O}vm|t x_Lqt =Q "OvDUy syx@ l}ORv = =

AirPassengers | iO=YD xirw-t %101 pmW

AirPassengers | iO =YD xirw-t Q=ov|oDU@ty %111 pmW

" CU= wt N (0 2) x C | = O}i x w Q m W=O Uw o U i v | U

C kw= = |a@ O \UwD xOW

" CU= 12 pmW q=@ = O

Cov(wt wt+k ) = E (wt ; )(wt+k ; )]

O}iU xiwv |oDU@tyOwN `@=D V}=tv %22 pmW

O}iU xiwv | Q_v |oDU@tyOwN `@=D V}=tv %32 pmW

k (t) = p Cov(xt xt+k ) = p 2 t 1

> for (t in 2:1000) xt] <{ xt - 1] + wt]

|iO=YD uORs o Q = =ov|oDU@tyOwN V}=tv %52 pmW

p=iD p=ta= =@ |iO=YD =