Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
25 views

Tutorial-Topic 2

This document provides instructions and questions for a time series analysis and forecasting tutorial. It includes: 1. Demonstrating that single exponential smoothing is more flexible than moving averages by allowing weights based on data importance. 2. Explaining that for a certain alpha value, single exponential smoothing approaches the naive model. 3. Defining a "benchmark" model and providing an example. 4. Fitting naive, single exponential smoothing, and linear regression trend models to data and generating one-step-ahead forecasts. 5. Writing Excel instructions to generate error terms for naive, single exponential, and Brown's exponential smoothing models.

Uploaded by

Aqim Azam
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
25 views

Tutorial-Topic 2

This document provides instructions and questions for a time series analysis and forecasting tutorial. It includes: 1. Demonstrating that single exponential smoothing is more flexible than moving averages by allowing weights based on data importance. 2. Explaining that for a certain alpha value, single exponential smoothing approaches the naive model. 3. Defining a "benchmark" model and providing an example. 4. Fitting naive, single exponential smoothing, and linear regression trend models to data and generating one-step-ahead forecasts. 5. Writing Excel instructions to generate error terms for naive, single exponential, and Brown's exponential smoothing models.

Uploaded by

Aqim Azam
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

TUTORIAL 2 STA 570/572 TIME SERIES ANALYSIS AND FORECASTING

1. “Single exponential smoothing model has the advantage over moving average in that it has the
flexibility to provide weights according to the importance of the data”. Discuss and demonstrate
the truth of this statement.

2. Explain that for a certain value of 𝛼 the single exponential smoothing approaches the naïve model.

3. What is a “benchmark” model? Explain and give an example.

4. Use the data below to fit the following models. The first eight (8) observations are to be used to
estimate the model. Then, using the estimated models, generate the one-step-ahead forecasts for
periods 1988 till 1990. The models are:

i. Naïve Forecast
ii. Single exponential smoothing
iii. Linear regression trend fitting

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
25 24 20 22 24 25 23 21 23 30 28

5. Write the necessary Excel instructions in the cells marked “?” in order to generate the error terms
for the models given below, assuming that the following table represents an Excel spreadsheet.
The bold alphabets indicate the columns. You are also to name the model used in each section.

𝑦𝑡
i. 𝐹𝑡+1 = (𝑦𝑡 𝑦 )
𝑡−1
Rows
A B C D E F
no.
1 t 𝑦𝑡 𝐹𝑡 𝜀1
2 1 𝑦1
3 2 𝑦2
4 3 𝑦3 ? ?
5 4 𝑦4

ii. 𝐹𝑡 = 𝛼𝑦𝑡 + (1 − 𝛼)𝐹𝑡−1 for 𝛼 = 0.8


Rows
A B C D E F
no.
1 t 𝑦𝑡 𝐹𝑡 𝜀1
2 1 𝑦1
3 2 𝑦2
4 3 𝑦3 ? ?
5 4 𝑦4
iii. 𝐴𝑡 = 𝛼𝑦𝑡 + (1 − 𝛼)(𝐴𝑡−1 + 𝑇𝑡−1 )
𝑇𝑡 = 𝛽(𝐴𝑡 − 𝐴𝑡−1 )(1 − 𝛽)𝑇𝑡−1
𝐹𝑡+𝑝 = 𝐴𝑡 + 𝑚 × 𝑇𝑡 for 𝛼 = 0.4, 𝛽 = 0.2 and m number of steps ahead forecast.
Rows
A B C D E F G
no.
1 t 𝑦𝑡 𝐴𝑡 𝑇1 𝐹𝑡+1 𝑒𝑡
2 1 𝑦1 ?
3 2 𝑦2 ? ? ?
4 3 𝑦3
5 4 𝑦4

6. How do you decide on the value of the parameter of an exponential model? Give two
examples that you can think of.

7. How do you determine the initial value of the exponential model?

8. Demonstrate the consequence of applying ‘small’ or ‘large’ parameter value in the


simple exponential model?

9. What is the implication of large 𝛼 in an exponential model?

10. For what use is the ‘hold-out’ data point?

11. In what way is the ARRES model different from other exponential models?

12. The application of statistical test procedure does not work in the exponential smoothing
modelling. Why?

You might also like