# Ti me Ser i es ARI MA Model s i n R

# Copyr i ght 2013 by Ani Kat chova

# i nst al l . packages( " t ser i es" )
l i br ar y( t ser i es)

mydat a<- r ead. csv( " C: / Economet r i cs/ Dat a/ t i meser i es_ppi . csv" )
at t ach( mydat a)

# Def i ni ng var i abl es
Y <- ppi
d. Y <- di f f ( Y)
t <- year qr t

# Descr i pt i ve st at i st i cs and pl ot t i ng t he dat a
summar y( Y)
summar y( d. Y)

pl ot ( t , Y)
pl ot ( d. Y)

# Di ckey- Ful l er t est f or var i abl e
adf . t est ( Y, al t er nat i ve=" st at i onar y" , k=0)
adf . t est ( Y, al t er nat i ve=" expl osi ve" , k=0)

summar y( l m( dppi ~ l ppi , na. act i on=na. omi t ) )
summar y( l m( dppi ~ l ppi + t r end, na. act i on=na. omi t ) )

# Augment ed Di ckey- Ful l er t est
adf . t est ( Y, al t er nat i ve=" st at i onar y" )

# DF and ADF t est s f or di f f er enced var i abl e
adf . t est ( d. Y, k=0)
adf . t est ( d. Y)


# ACF and PACF
acf ( Y)
pacf ( Y)

acf ( d. Y)
pacf ( d. Y)

# ARI MA( 1, 0, 0) or AR( 1)
ar i ma( Y, or der = c( 1, 0, 0) )

# ARI MA( 2, 0, 0) or AR( 2)
ar i ma( Y, or der = c( 2, 0, 0) )

# ARI MA( 0, 0, 1) or MA( 1)
ar i ma( Y, or der = c( 0, 0, 1) )

# ARI MA( 1, 0, 1) or AR( 1) MA( 1)
ar i ma( Y, or der = c( 1, 0, 1) )

# ARI MA on di f f er enced var i abl e
# ARI MA( 1, 1, 0)
ar i ma( d. Y, or der = c( 1, 0, 0) )

# ARI MA( 0, 1, 1)
ar i ma( d. Y, or der = c( 0, 0, 1) )

# ARI MA( 1, 1, 1)
ar i ma( d. Y, or der = c( 1, 0, 1) )

# ARI MA( 1, 1, 3)
ar i ma( d. Y, or der = c( 1, 0, 3) )

# ARI MA( 2, 1, 3)
ar i ma( d. Y, or der = c( 2, 0, 3) )


# ARI MA( 1, 0, 1) f or ecast i ng
mydat a. ar i ma101 <- ar i ma( Y, or der = c( 1, 0, 1) )
mydat a. pr ed1 <- pr edi ct ( mydat a. ar i ma101, n. ahead=100)
pl ot ( Y)
l i nes( mydat a. pr ed1$pr ed, col =" bl ue" )
l i nes( mydat a. pr ed1$pr ed+2*mydat a. pr ed1$se, col =" r ed" )
l i nes( mydat a. pr ed1$pr ed- 2*mydat a. pr ed1$se, col =" r ed" )

# ARI MA( 1, 1, 1) f or ecast i ng
mydat a. ar i ma111 <- ar i ma( d. Y, or der = c( 1, 0, 1) )
mydat a. pr ed1 <- pr edi ct ( mydat a. ar i ma111, n. ahead=100)
pl ot ( d. Y)
l i nes( mydat a. pr ed1$pr ed, col =" bl ue" )
l i nes( mydat a. pr ed1$pr ed+2*mydat a. pr ed1$se, col =" r ed" )
l i nes( mydat a. pr ed1$pr ed- 2*mydat a. pr ed1$se, col =" r ed" )


> # Ti me Ser i es ARI MA Model s i n R
> # Copyr i ght 2013 by Ani Kat chova
>
> # i nst al l . packages( " t ser i es" )
> l i br ar y( t ser i es)

t ser i es ver si on: 0. 10- 32

t ser i es i s a package f or t i me ser i es anal ysi s and comput at i onal f i nance.

See l i br ar y( hel p=" t ser i es" ) f or det ai l s.

>
> mydat a<- r ead. csv( " C: / Economet r i cs/ Dat a/ t i meser i es_ppi . csv" )
> at t ach( mydat a)
>
> # Def i ni ng var i abl es
> Y <- ppi
> d. Y <- di f f ( Y)
> t <- year qr t
>
> # Descr i pt i ve st at i st i cs and pl ot t i ng t he dat a
> summar y( Y)
Mi n. 1st Qu. Medi an Mean 3r d Qu. Max.
25. 24 29. 66 77. 00 64. 68 93. 27 110. 40
> summar y( d. Y)
Mi n. 1st Qu. Medi an Mean 3r d Qu. Max.
- 3. 2100 0. 0300 0. 3000 0. 4643 0. 9625 3. 0800
>
> pl ot ( t , Y)
> pl ot ( d. Y)
>
> # Di ckey- Ful l er t est f or var i abl e
> adf . t est ( Y, al t er nat i ve=" st at i onar y" , k=0)

Augment ed Di ckey- Ful l er Test

dat a: Y
Di ckey- Ful l er = - 0. 7931, Lag or der = 0, p- val ue = 0. 9604
al t er nat i ve hypot hesi s: st at i onar y

> adf . t est ( Y, al t er nat i ve=" expl osi ve" , k=0)

Augment ed Di ckey- Ful l er Test

dat a: Y
Di ckey- Ful l er = - 0. 7931, Lag or der = 0, p- val ue = 0. 03964
al t er nat i ve hypot hesi s: expl osi ve

>
> summar y( l m( dppi ~ l ppi , na. act i on=na. omi t ) )

Cal l :
l m( f or mul a = dppi ~ l ppi , na. act i on = na. omi t )

Resi dual s:
Mi n 1Q Medi an 3Q Max
- 3. 6484 - 0. 4507 - 0. 1697 0. 5135 2. 6168

Coef f i ci ent s:
Est i mat e St d. Er r or t val ue Pr ( >| t | )
( I nt er cept ) 0. 5035691 0. 1682689 2. 993 0. 00319 **
l ppi - 0. 0006095 0. 0023653 - 0. 258 0. 79697
- - -
Si gni f . codes: 0 *** 0. 001 ** 0. 01 * 0. 05 . 0. 1 1

Resi dual st andar d er r or : 0. 9233 on 166 degr ees of f r eedom
( 1 obser vat i on del et ed due t o mi ssi ngness)
Mul t i pl e R- squar ed: 0. 0003999, Adj ust ed R- squar ed: - 0. 005622
F- st at i st i c: 0. 0664 on 1 and 166 DF, p- val ue: 0. 797

> summar y( l m( dppi ~ l ppi + t r end, na. act i on=na. omi t ) )

Cal l :
l m( f or mul a = dppi ~ l ppi + t r end, na. act i on = na. omi t )

Resi dual s:
Mi n 1Q Medi an 3Q Max
- 3. 7271 - 0. 4264 - 0. 2088 0. 5155 2. 6533

Coef f i ci ent s:
Est i mat e St d. Er r or t val ue Pr ( >| t | )
( I nt er cept ) 0. 581139 0. 197371 2. 944 0. 0037 **
l ppi - 0. 008389 0. 010578 - 0. 793 0. 4289
t r end 0. 004957 0. 006569 0. 755 0. 4516
- - -
Si gni f . codes: 0 *** 0. 001 ** 0. 01 * 0. 05 . 0. 1 1

Resi dual st andar d er r or : 0. 9245 on 165 degr ees of f r eedom
( 1 obser vat i on del et ed due t o mi ssi ngness)
Mul t i pl e R- squar ed: 0. 003838, Adj ust ed R- squar ed: - 0. 008237
F- st at i st i c: 0. 3178 on 2 and 165 DF, p- val ue: 0. 7282

>
> # Augment ed Di ckey- Ful l er t est
> adf . t est ( Y, al t er nat i ve=" st at i onar y" )

Augment ed Di ckey- Ful l er Test

dat a: Y
Di ckey- Ful l er = - 1. 3857, Lag or der = 5, p- val ue = 0. 8327
al t er nat i ve hypot hesi s: st at i onar y

>
> # DF and ADF t est s f or di f f er enced var i abl e
> adf . t est ( d. Y, k=0)

Augment ed Di ckey- Ful l er Test

dat a: d. Y
Di ckey- Ful l er = - 6. 8398, Lag or der = 0, p- val ue = 0. 01
al t er nat i ve hypot hesi s: st at i onar y

War ni ng message:
I n adf . t est ( d. Y, k = 0) : p- val ue smal l er t han pr i nt ed p- val ue
> adf . t est ( d. Y)

Augment ed Di ckey- Ful l er Test

dat a: d. Y
Di ckey- Ful l er = - 3. 2459, Lag or der = 5, p- val ue = 0. 08252
al t er nat i ve hypot hesi s: st at i onar y

>
>
> # ACF and PACF
> acf ( Y)
> pacf ( Y)
>
> acf ( d. Y)
> pacf ( d. Y)
>
> # ARI MA( 1, 0, 0) or AR( 1)
> ar i ma( Y, or der = c( 1, 0, 0) )

Cal l :
ar i ma( x = Y, or der = c( 1, 0, 0) )

Coef f i ci ent s:
ar 1 i nt er cept
0. 9996 64. 522
s. e. 0. 0005 38. 002

si gma^2 est i mat ed as 1. 058: l og l i kel i hood = - 248. 2, ai c = 502. 4
>
> # ARI MA( 2, 0, 0) or AR( 2)
> ar i ma( Y, or der = c( 2, 0, 0) )

Cal l :
ar i ma( x = Y, or der = c( 2, 0, 0) )

Coef f i ci ent s:
ar 1 ar 2 i nt er cept
1. 6474 - 0. 6475 230. 406
s. e. 0. 0003 0. 0003 NaN

si gma^2 est i mat ed as 0. 6061: l og l i kel i hood = - 198. 27, ai c = 404. 54
War ni ng messages:
1: I n l og( s2) : NaNs pr oduced
2: I n sqr t ( di ag( x$var . coef ) ) : NaNs pr oduced
>
> # ARI MA( 0, 0, 1) or MA( 1)
> ar i ma( Y, or der = c( 0, 0, 1) )

Cal l :
ar i ma( x = Y, or der = c( 0, 0, 1) )

Coef f i ci ent s:
ma1 i nt er cept
1. 0000 64. 6863
s. e. 0. 0182 2. 3345

si gma^2 est i mat ed as 231. 6: l og l i kel i hood = - 702. 48, ai c = 1410. 96
>
> # ARI MA( 1, 0, 1) or AR( 1) MA( 1)
> ar i ma( Y, or der = c( 1, 0, 1) )

Cal l :
ar i ma( x = Y, or der = c( 1, 0, 1) )

Coef f i ci ent s:
ar 1 ma1 i nt er cept
0. 9994 0. 5337 64. 6936
s. e. 0. 0008 0. 0537 37. 6621

si gma^2 est i mat ed as 0. 7233: l og l i kel i hood = - 216. 41, ai c = 440. 82
>
> # ARI MA on di f f er enced var i abl e
> # ARI MA( 1, 1, 0)
> ar i ma( d. Y, or der = c( 1, 0, 0) )

Cal l :
ar i ma( x = d. Y, or der = c( 1, 0, 0) )

Coef f i ci ent s:
ar 1 i nt er cept
0. 5522 0. 4557
s. e. 0. 0640 0. 1307

si gma^2 est i mat ed as 0. 5841: l og l i kel i hood = - 193. 39, ai c = 392. 79
>
> # ARI MA( 0, 1, 1)
> ar i ma( d. Y, or der = c( 0, 0, 1) )

Cal l :
ar i ma( x = d. Y, or der = c( 0, 0, 1) )

Coef f i ci ent s:
ma1 i nt er cept
0. 4872 0. 4654
s. e. 0. 0579 0. 0908

si gma^2 est i mat ed as 0. 6284: l og l i kel i hood = - 199. 5, ai c = 404. 99
>
> # ARI MA( 1, 1, 1)
> ar i ma( d. Y, or der = c( 1, 0, 1) )

Cal l :
ar i ma( x = d. Y, or der = c( 1, 0, 1) )

Coef f i ci ent s:
ar 1 ma1 i nt er cept
0. 7245 - 0. 2547 0. 4397
s. e. 0. 1152 0. 1682 0. 1576

si gma^2 est i mat ed as 0. 5783: l og l i kel i hood = - 192. 59, ai c = 393. 17
>
> # ARI MA( 1, 1, 3)
> ar i ma( d. Y, or der = c( 1, 0, 3) )

Cal l :
ar i ma( x = d. Y, or der = c( 1, 0, 3) )

Coef f i ci ent s:
ar 1 ma1 ma2 ma3 i nt er cept
0. 7334 - 0. 241 - 0. 1082 0. 1217 0. 4324
s. e. 0. 1242 0. 142 0. 0970 0. 0800 0. 1664

si gma^2 est i mat ed as 0. 5638: l og l i kel i hood = - 190. 48, ai c = 392. 97
>
> # ARI MA( 2, 1, 3)
> ar i ma( d. Y, or der = c( 2, 0, 3) )

Cal l :
ar i ma( x = d. Y, or der = c( 2, 0, 3) )

Coef f i ci ent s:
ar 1 ar 2 ma1 ma2 ma3 i nt er cept
1. 5191 - 0. 7084 - 1. 0502 0. 2100 0. 3179 0. 4405
s. e. 0. 2253 0. 1589 0. 2103 0. 1314 0. 1036 0. 1438

si gma^2 est i mat ed as 0. 5474: l og l i kel i hood = - 188. 22, ai c = 390. 44
>
>
> # ARI MA( 1, 0, 1) f or ecast i ng
> mydat a. ar i ma101 <- ar i ma( Y, or der = c( 1, 0, 1) )
> mydat a. pr ed1 <- pr edi ct ( mydat a. ar i ma101, n. ahead=100)
> pl ot ( Y)
> l i nes( mydat a. pr ed1$pr ed, col =" bl ue" )
> l i nes( mydat a. pr ed1$pr ed+2*mydat a. pr ed1$se, col =" r ed" )
> l i nes( mydat a. pr ed1$pr ed- 2*mydat a. pr ed1$se, col =" r ed" )
>
> # ARI MA( 1, 1, 1) f or ecast i ng
> mydat a. ar i ma111 <- ar i ma( d. Y, or der = c( 1, 0, 1) )
> mydat a. pr ed1 <- pr edi ct ( mydat a. ar i ma111, n. ahead=100)
> pl ot ( d. Y)
> l i nes( mydat a. pr ed1$pr ed, col =" bl ue" )
> l i nes( mydat a. pr ed1$pr ed+2*mydat a. pr ed1$se, col =" r ed" )
> l i nes( mydat a. pr ed1$pr ed- 2*mydat a. pr ed1$se, col =" r ed" )
>