Scribd
Upload
8 views11 pages

Tutorial 10

Uploaded by

salmarefaie
Copyright
© © All Rights Reserved
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
8 views11 pages

Tutorial 10

Uploaded by

salmarefaie
Copyright
© © All Rights Reserved
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/ 11

Topic 11: Time Series Forecasting Tutorial 10

Week 11
Time Series Definition
• Any data that is observed over time is called time series data.
• The aim is to analyse the past behaviour of a variable in order to predict its future
behaviour.
Types of TS data:
• Stationary Data: TS data that exhibits no significant upward or downward trend
overtime.
• Non-Stationary Data: TS data that exhibits a significant upward or downward
trend overtime.
• Seasonal Data: TS data that repeat patterns at regular intervals (seasons) over
time.
Time Series Approaches

A. Extrapolation (Smoothing) Techniques.


Used for Stationary Data, with no seasonality.
✓ Moving Average.
✓ Weighted Moving Average.
✓ Exponential Smoothing

B. Modelling (Regression) Techniques.


Measuring Accuracy
Question 1
Consider the following TS data given below:

Year Sales
1 283
2 288
3 336
4 388
5 406
6 412
7 416
8 435
9 428
10 435
Required:

1. Prepare a line graph of these data. Do the data appear to be stationary or


nonstationary?
2. Compute the two-period Moving average and four-period Moving average
predictions for the data set.
3. Compute forecasts for the next two years using the two-period and four-period
MA techniques.
4. Using MSE, which MA technique is more accurate?
1. The Line Graph:
A
Sales
n 500

s 450

400
w 350

e 300

250
r 200

150

100

50

0
1 2 3 4 5 6 7 8 9 10

Comment:
The data appears to be stationary since it does not exhibit a significant upward or downward trend overtime.
2. The two-period Moving average and four-period Moving average predictions:

A Year Sales MA(2) MA(4)


n 1 283 -- --
s 2 288 -- --
w
3 336 285.5 --
e
4 388 312 --
r
5 406 362 323.75

6 412 397 354.5

7 416 409 385.5

8 435 414 405.5

9 428 425.5 417.25

10 435 431.5 422.75


3. Forecasts for the next two years using the two-period and four-period MA techniques.

A
n Year Sales MA(2) MA(4)
s 1 283
2 288
w 3 336 285.5
e 4 388 312

r 5 406 362 323.75


6 412 397 354.5
7 416 409 385.5
8 435 414 405.5
9 428 425.5 417.25
10 435 431.5 422.75

11 431.5 428.5
12 431.5 428.5
4. MSE calculation for MA(2) and MA(4).

A Error |Error| Error |Error|


Year Sales MA(2) (Error)^2 MA(4) (Error)^2
(yi - ŷi) |yi - ŷi | (yi - ŷi) |yi - ŷi |
n 1 283 -- -- -- -- -- -- -- --
s 2 288 -- -- -- -- -- -- -- --
3 336 285.5 50.5 50.5 2,550.25 -- -- -- --
w
4 388 312 76 76 5,776 -- -- -- --
e 5 406 362 44 44 1,936 323.75 82.25 82.25 6,765.06
r 6 412 397 15 15 225 354.5 57.5 57.5 3,306.25
7 416 409 7 7 49 385.5 30.5 30.5 930.25
8 435 414 21 21 441 405.5 29.5 29.5 870.25
9 428 425.5 2.5 2.5 6.25 417.25 10.75 10.75 115.56
10 435 431.5 3.5 3.5 12.25 422.75 12.25 12.25 150.06
Total 10,995.75 12,137.44
n 8 6
MSE 1,374.47 2,022.91
Since the MSE of MA(2) is lower than the MSE of MA(4); 1,374.47 < 2,022.91.
Therefore, MA(2) provides more accurate forecasts than MA(4). The smaller MSE, the better.

You might also like