Scribd
Upload
68 views3 pages

STAT 248: EDA & Stationarity Answers Exercise Lab 3: 1 Theoretical Problems

1. The time series xt is non-stationary as its mean depends on time and is not constant. 2. The process yt = xt - xt-1 is stationary as its mean is constant, it has a finite second moment, and its autocovariance function only depends on the lag h and not time t. 3. The mean of the moving average vt is β1 + β2t and its autocovariance function simplifies to 2σZ2q+1-h(2q+1)2 if the lag |h| is less than or equal to 2q, and 0 otherwise.

Uploaded by

Kaled Abode
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
68 views3 pages

STAT 248: EDA & Stationarity Answers Exercise Lab 3: 1 Theoretical Problems

1. The time series xt is non-stationary as its mean depends on time and is not constant. 2. The process yt = xt - xt-1 is stationary as its mean is constant, it has a finite second moment, and its autocovariance function only depends on the lag h and not time t. 3. The mean of the moving average vt is β1 + β2t and its autocovariance function simplifies to 2σZ2q+1-h(2q+1)2 if the lag |h| is less than or equal to 2q, and 0 otherwise.

Uploaded by

Kaled Abode
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

STAT 248: EDA & Stationarity Answers Exercise Lab 3

GSI: Gido van de Ven September 17th, 2010

Theoretical Problems

Consider the time series

xt = 1 + 2 t + Zt

(t = 1, 2, 3, ...)

(1)

2 where 1 and 2 are known constants and Zt is a white noise process with variance Z .

1. Determine whether xt is stationary. E(xt ) = E(1 + 2 t + Zt ) = 1 + 2 t using E(Zt ) = 0

As the mean function depends on time and is not constant, xt is non-stationary. 2. Show that the process yt = xt xt1 is stationary. We need to check the three requirements of a stationary process given in the handouts.

E(yt ) = E(xt xt1 ) = E(1 + 2 t + Zt (1 + 2 (t 1) + Zt1 )) = 2 where we used that E(Zt ) = 0 t. So the mean function is constant. Also, yt has a nite second moment since it is a sum of two random variables with nite second moments. The only thing left to verify is that the auto covariance function is only a function of the lag h, and does not depend on time t.

(t, t + h) = Cov(yt , yt+h ) = E[yt yt+h ] E[yt ]E[yt+h ] Working with the rst part of (2) we have the following result:

(2)

E[yt yt+h ] = E[(xt xt1 )(xt+h xt+h1 )] = E[(Zt + 2 Zt1 )(Zt+h + 2 Zt+h1 )]
2 = E[Zt Zt+h ] E[Zt Zt+h1 ] + 2 + E[Zt1 Zt+h1 ]

2 2 where we used thatE(Zt1 Zt2 ) = 0 t1 = t2 . Also using E(Zt ) = Z gives:

2 2 2 + 2Z 2 2 E[yt yt+h ] = Z 2 2 2 Working with the second part of (2) gives:

if h = 0 if h = 1 else

2 E[yt ]E[yt+h ] = E[xt xt1 ]E[xt+h xt+h1 ] = E[yt ]E[yt+h ] = 2

So: 2 2Z 2 Z Cov(yt , yt+h ) = 0 if h = 0 if h = 1 else

Thus the auto covariance function does not depend on time t. Therefore, yt is stationary. 3. Show that the mean of the moving average 1 vt = 2q + 1
q

xtj
j=q

(3)

is 1 + 2 t, and give a simplied expression for the autocovariance function of vt . 1 2q + 1


q

E(vt ) =

j=q

1 E(xtj ) = 2q + 1
q

E(1 + 2 (t j) + Ztj )
j=q

= 1 +

2 2q + 1

(t j) = 1 + 2 t +
j=q

2 2q + 1

j = 1 + 2 t
j=q

For the autocovariance function it is useful to rst calculate:


2 2 2 1 + 21 2 t + 2 t2 + Z if j = 0 2 + 2 t + j + 2 t(t j) else 1 1 2 1 2 2

E(xt xtj ) = E[(1 + 2 t + Zt )(1 + 2 (t j) + Ztj )] Using this, we can calculate:


q q

E[vt vt+h ] =

1 E[ (2q + 1)2

xtj

j=q

2 = 1 + 21 2

1 (2q + 1)2
q

j=q q q

1 xt+hj ] = (2q + 1)2 (t j) + 1 2

E[xtj xt+hi ]
j=q i=q q q

... +
2 1

2 2

1 (2q + 1)2

j=q i=q q

1 (2q + 1)2

(h + j i) + ...
j=q i=q 2 Z

[(t j)((t j) + (h + j i))] +


j=q i=q 2 2 t(t 2 Z

1 (2q + 1)2

1{h+ji=0}
j=q i=q

+ 21 2 t + 1 2 h +

+ h) +

1 (2q + 1)2

1{h+ji=0}
j=q i=q

Also:

2 2 E[vt ]E[vt+h ] = (1 + 2 t)(1 + 2 (t + h)) = 1 + 21 2 t + 1 2 h + 2 t(t + h)

Using above results, we can simplify the expression for the autocovariance function of vt as:
q q

Cov(vt , vt+h ) = E[vt vt+h ] E[vt ]E[vt+h ] =

2 Z

1 (2q + 1)2

1{h+ji=0} =
j=q i=q

2 Z 2q+1h (2q+1)2 0

if |h| 2q else

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy