stovhastic processes

stovhastic processes

Instructions:
• You MUST comply with Academic Integrity as indicated on the electronic submission. Please note that this is an INDIVIDUAL assignment, not a group assignment. Inappropriate collaboration will be penalized.
• Any data files will be attached to the link provided on the STAT3702/STAT8702 FLOsite under Week 8/Assessment/Assignment 2.
• Your submission should consist of at most 3 files in PDF format with a total size no bigger than 50 MB.
• You can update your submission for unlimited number of times before the due date.
• Refer to the “Statement of Assessment Methods” pdf document on FLO regarding late assignment penalties.
• Medical extension or extension due to compassionate ground may be granted. Only applications with legitimate reasons will be considered.
• Keep a copy of the submission yourself.
Writing Up Your Assignment….
• Answer all questions in this assignment. Questions should be answered in the order they appear.
• MS-Word (or other typesetting software of your choice) may be used in preparing your assignment submission whenever appropriate, which would then being converted into pdf.
• SPSS or R may be used to assist with calculations. Answers must be written in clear English sentences with all appropriate working and/or supporting computer output shown. Raw computer output without explanatory text is unacceptable.
• All graphs and plots must be constructed properly using either SPSS or R. Free-hand sketching will be awarded with no marks.
• Do not include any graphs, plots or tables in the appendix or attachments. Instead, graphs, plots and tables should be resized properly and being included within the appropriate question.
• All workings and intermediate answers must be clearly shown.
STAT3702/8702 Assignments– Sem 2, 2017
page 2
_______________________________________________________________
Note: In all problems, we assume that } { t ω
is a white noise with mean zero and a constant
variance 2ω
σ .
Question 1 [Total: 10 marks]
(a) From a series of length n=100, we calculate the sample autocorrelations as r1 = -0.49, r2 = 0.31,
r3 = -0.21, r4 = 0.11 and |rk | ≤ 0.09 for k > 4. On the basis of this information alone, what
ARIMA model would we tentatively specify for the series ? [5 marks]
(b) Consider an MA(1) process for which it is known that the mean is zero. Based on a series of
length 3, we observe
x1 = 0, x2 = -1, x3 = 0.5
(i) Show that the conditional least squares estimate of θ is 0.5 [3 marks]
(ii) Find an estimate of the noise variance σw
2 [2 marks]
Question 2 [Total: 10 marks]
Suppose that the process {xt } develops according to
xt = xt −4 +ωt , for t > 4,
with xt =ωt for t=1,2,3,4.
Find
(a) the variance function for {xt }, [4 marks]
(b) autocorrelation function for {xt }, [4 marks]
(c) identify the model as a certain seasonal ARIMA model. [2 marks]
Question 3 [Total: 15 marks]
Based on quarterly data, a seasonal model of the form
−4 1 −1 2 −2 = + − − t t t t t x x ω θ ω θ ω
has been fitted to a certain time series.
(a) Find the first four Ψ weights for this model analytically. [5 marks]
(b) Suppose that 0.5, 0.25, 1 1 2 = = − = ω θ θ σ . Find the forecasts for the next four quarters if data for
the last four quarters are
STAT3702/8702 Assignments– Sem 2, 2017
page 3
Quarter I II III IV
Series 25 20 25 40
Residual 2 1 2 3
[5 marks]
(c) Find the 95% prediction intervals for the forecasts in part (b). [5 marks]
Question 4 [Total: 15 marks]
The dataset airpass2017.sav (airpass.dat) shows total monthly air-passenger-miles within the
United States from January 1960 through December 1977.
(a) Specify and estimate the parameters for at least two tentative ARIMA models for the period of
January 1960 to October 1977. Comment on each statistical output to support your tentative
models. [4 marks]
(b) Perform diagnostic checking on the models estimated in part (a), repeat the process in (a) until
you obtain a final model. [4 marks]
(c) Write down your final model in its mathematical form. Interpret the model. [3 marks]
(d) Use the model in (c) to forecast the total monthly air-passenger-miles within the United States in
November and December 1977 respectively, manually and using SPSS or R, including its 95%
confidence interval. [4 marks]
Question 5 [STAT8702 ONLY] [Total: 10 marks]
(a) Consider the ARIMA model
xt =ωt + Θωt −2
Show that the series is invertible for Θ < 1, and find the coefficients π in the representation
= Σ

=

k 0
ωt π k xt k [5 marks]
(b) Consider the ARCH(1) model with a constant mean:
( )
2 1,t-1 ( 2 )
0 1 1
2
1
; | y ~ N 0,
~ N 0, 1
0
t t t t
t t t
a a
y a a
σ α α σ
μ α
α


= +
= +
Show that the model, 2
0 1 −1 = + t t t y y α α ε , where εt
~ N(0,1) and is an independent white noise
series, is equivalent to an ARCH(1) model with 0 mean. [5 marks]
End of Assignment 2

Click here to request for this assignment help