Professional Documents
Culture Documents
L Dhliwayo
Department of Statistics
University of Zimbabwe
Contents
1 Models for Non-Stationary Time Series 1
1.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Models for Non-Stationary Series: ARIMA(p,d,q) Processes . . . . . . . 1
1.3 Statinarity Through Differencing . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Random Walk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Seasonal ARMA Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 Non Stationary Seasonal Processes . . . . . . . . . . . . . . . . . . . . . 5
i
Unit 1
Models for Non-Stationary Time
Series
Unit Objectives
1. ,
2.
3.
4.
5.
6.
7.
8.
There are various transformations that we use to make a time series stationary. Some
of them are:
1. differencing
1
2 Control Charts for Variables
2. log transformation
4. arcsine transformation
where
d
X d
∆d Xt = (−1)j B j Xt (1.2)
j
j=0
Solution 1.1
Xt = 32 Xt−1 − 12 Xt−2 + at
⇒ (1 − 23 B + 12 B 2 )Xt = at
⇒ φ(B) = (1 − 23 B + 12 B 2 ) = 0 ⇒ B = 1 or B = 2
⇒ Xt is not stationary since one of the roots lie on the unit circle.
⇒ Zt = 12 Zt−1 + at
⇒ (1 − 21 )Zt = at
⇒ φ(B) = (1 − 12 ) = 0 ⇒ B=2
Control Charts for Variables 3
Solution 1.2 1.
E(Xt ) = E(5 + 2t + at )
= 5 + 2t + E(at )
= 5 + 2t + µ
2.
Z1 = a1
Z2 = a1 + a2
Z3 = a1 + a2 + a3
..
.
Zt = a1 + a2 + . . . + at
Zt = Zt−1 + at (1.3)
E(Zt ) = E(a1 + a2 + . . . + at ) = tµ
4 Control Charts for Variables
V ar(Zt ) = V ar(a1 + a2 + . . . + at )
For 1 ≤ t ≤ s
Therefore
r
tσa2 t
ρt,s = p = f or 1 ≤ t ≤ s (1.4)
tσa2 × sσa2 s
Definition 1.1 A process {Zt } is called a seasonal ARMA process of non seasonal
order p, q and seasonal component P, Q and seasonanility of order S if Zt satisfies:
Φ(B) = 1 − Φ1 B S − Φ2 B 2S − ΦP B SP
Θ(B) = 1 − Θ1 B S − Θ2 B 2S − ΘQ B SQ
φ(B) = 1 − φ1 B − φ2 B 2 − φp B p
θ(B) = 1 − θ1 B − θ2 B 2 − θq B q
Control Charts for Variables 5
1. ARMA(0,0)(0,2)12
2. ARMA(0,0)(2,0)4
2. ARMA(0,0)(2,0)4
D
j D
X
∆D
S Xt = (−1) B jS Xt (1.6)
j
j=0
Definition 1.3 A process {Zt } is called a seasonal ARIMA process of non seasonal
components p, d, q and seasonal component P, D, Q and seasonal component of order S
if Zt satisfies:
Φ(B)φ(B)∆d ∆D
S Zt = Θ(B)θ(B)at (1.8)
where
Φ(B) = 1 − Φ1 B S − Φ2 B 2S − ΦP B SP
Θ(B) = 1 − Θ1 B S − Θ2 B 2S − ΘQ B SQ
φ(B) = 1 − φ1 B − φ2 B 2 − φp B p
θ(B) = 1 − θ1 B − θ2 B 2 − θq B q
1. ARMA(0,1,1)(0,1,2)12
2. ARMA(0,1,0)(2,1,0)4