EQC7006: Time Series Analysis 2018
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
ACADEMIC SESSION 2017/2018: SEMESTER II
EQC7006: Time Series Analysis
1. The following table gives a quarterly time series data and the results of classical decomposition using multiplicative decomposition approach.
a) Compute seasonal indices (1 and 2 ), normalized seasonal indices ( 1 , 2 , 3 and 4 ), and revised irregular component 3, 4, 5 and 6 values. (6 marks)
b) Compute seasonally adjusted series ( 1, 2, 3 and 4 ) using normalized obtained in (a). (2 marks)
c) Give an example of situation that requires us to analyse seasonally adjusted series instead of the seasonally unadjusted series. (0.5 mark)
(Total: 8.5 marks)
2. The following table gives a quarterly time series data and the results of Holt-Winters multiplicative exponential smoothing approach. ( = = = 0.3).
a) Compute the initial values 4 , 4 , 1 , 2 , 3 , and 4 . (4 marks)
b) Compute the updated values 5 , 5 , 5 , and 6 . (4 marks)
c) Compute forecast values 9 and 12 . (2 marks)
d) Sketch an example of time series data that has multiplicative trend. (0.5 mark)
e) Sketch an example of time series data that has damped trend. (0.5 mark)
f) Sketch an example of time series data that has additive seasonality.(0.5 mark)
g) Sketch an example of time series data that has multiplicative seasonality. (0.5 mark)
(Total: 12 marks)
3. The following tables give estimated coefficients for three linear regression models. Y denotes the monthly electricity demand (in megawatts) series. T represents time trend (T = 1, 2, 3, …, 96). Di’s are seasonal dummies where D1=1 for January and 0 otherwise, and D2, D3, D4, D5, D6, D7, D8, D9, D10, D11 and D12 are similarly defined for February, March until December, respectively.
Model 1
Dependent Variable: Y
Method: Least Squares
Sample: 2009M01 2016M12
Included observations: 96
Model 2
Dependent Variable: Y
Method: Least Squares
Sample: 2009M01 2016M12
Included observations: 96
a) Interpret the estimated slope coefficients for T, D1 and D2 of Model 1. (3 marks)
b) Perform a seasonal variation test at 5% significance level. State the null and alternative hypotheses of the test. (3.5 marks)
c) Perform a test of equal seasonal effects in July and October at 5% significance level. State the null and alternative hypotheses of the test. (3.5 marks)
d) Assuming that the estimated Model 1 is an adequate model, compute forecasts for January and February 2017. Given that forecasts standard errors for January and February 2017 are 316.4942 and 316.4942, respectively, compute 95% forecast intervals. (4 marks)
e) Inspect the following correlogram of residuals for estimated Model 1 and determine whether the estimated model is an adequate model or not. If not, propose an augmented model and provide justifications. (3 marks)
4. Refer to the time series data and results given in Question 3.
a) Explain what should be done for the series to achieve stationarity. (0.5 mark)
b) Analyze the following result of unit root tests on seasonally differenced series(1 − L12)yt . Decide whether further differencing is required or not. (2 marks)
c) Given below are estimation results of two different models with their correlogram of residuals. Select a model based on the value of Schwarz Information Criterion (SIC) and write the model in terms of the backshift operator and without backshift operator. Then select a model based on the value of Akaike Information Criterion (AIC) and write the model in terms of the backshift operator and without backshift operator. (5 marks)
d) Explain what does parsimonious model mean? Which model should be selected based on parsimonious model? Does the selected model is an adequate model? Explain. (3 marks)
e) Table A below gives the last 24 observations of the series (January 2015 to December 2016). Table B on the other hand gives the last 24 residuals generated by Model 2. Compute forecasts for January and February of 2017.
2022-06-15