CHAPTER 5

TIME SERIES

5.1 INTRODUCTION TIME SERIES

Time series data are usually characterized by certain unique attributes. The presence of this
attributes in the data series may influences the decision to select or not to select the
particular model. Therefore, the process of determining and developing the most suitable
model to generate the forecast should begin with developing and understanding the
attribute, for example their behavior over time and their effect on the selected model.

5.1.1 Definitions of time series data

–Definition- a set of data collected or arranged in a sequence of order over a successive

equal increment of time.

–Time period may be given in terms of years or months or days or hours and so forth.

–Example - a country’s annual gross domestic product taken over 30 year period, the
monthly output from a factory taken from January 2000 till December 2016, etc.

5.1.2 Components of time series

–Four main component types: trend component, cyclical component, seasonal component
and irregular component.

–Classical decomposition time series – a process of analyzing those components in a

systematic manner.

–Symbols will be used to represent the components in time series:

Tt – trend component in the time period t
Ct – cyclical component in the time period t
It – irregular component in the time period t
St – seasonal component in the time period t

81
5.2 Trend

The trend component of a time series is defined as the long term general movement where
the value of the variable tends to increase or decrease over a long period of time (more than
10 years).

Example 48

-An increasing number of HIV cases in Malaysia from year to year.

-An increasing number of road accidents’ death from year to year.

-A declining rate of babies’ death from year to year due to modern medicines.

-An increasing number

5.3 Cyclical Variations

This component repeats its pattern over a period of time (2-10 years). The most common
example of cyclical variation is the business cycle which is usually caused by the policies of
the company, the host country and consumers’ taste.

Example 49

-A declining rate of share due to economic recession.

-A decreasing employment rate due to economic downturn.

-An increasing rate of retrenchment among factory workers due to economic recession.

82
5.4 Seasonal Variations

Seasonal variations involve patterns of change which are repeated from year to year.
Usually they are caused by weather and holidays that occur during the course of a year.

Example 50

-Increasing sales of bus tickets during the festive seasons.

-Increasing sales of school uniforms during the school holidays.

-Increasing number of people in KLIA during the Hajj season.

-Increasing number of traffics on the road from 7.00 am – 9.00 am and from 4.30 pm – 7.00
pm.

5.5 Irregular Variations

It describes the movements of a variable which is completely unpredictable. This is usually

caused by catastrophes. Examples:
-earthquake, Tsunami.
-A decreasing number of tourists flying to New York due to the attack on World Trade Centre
on the 11th September 2001.
-A loss of millions of dollars due to fires which caught the company last week.
-A declining rate of share prices of New York Stock Exchange due to the attack on the World
Trade Centre.
-A loss of properties due to floods in Kuantan last January 2002.

83
5.6 Trend Analysis
•Trend can be described using a histogram, where the independent variable (x- axis) is time
and dependent variable (y-axis) is the observed variable.

•Other methods include the least squares and moving average methods.

•We use centered moving average when data are monthly, quarterly, or when the period is
an even number of terms.

•Simple moving average is used when data are such that the period is an odd number of
terms, for example five days a week, three shifts a day.

•The purpose of studying trend analysis is to predict the future based on the past.

5.6.1 Estimation Trend Using Moving Average Method

FOR ODD NUMBER DATA SERIES.

3-period Moving Average

Time (t) yt Mt

1 y1 -

y1  y 2  y 3
2 y2 M1 
3

y 2  y3  y 4
3 y3 M2 
3

y3  y 4  y5
4 y4 M3 
3

y 4  y5  y6
5 y5 M4 
3

6 y6 -

84
Example 51

The following figures relate to the sales of certain item at Shop X provided operating 24
hours a day.

Sales (‘ 000)
Day
Morning (M) Afternoon (A) Evening (E)
Monday 25 23 26
Tuesday 24 25 26
Wednesday 24 23 25
Thursday 23 22 27
Friday 24 22 26

I. Calculate the trend using the moving average method.

II. Forecast the trend of sales for Morning (M) and Afternoon (A) for Saturday.

FOR EVEN DATA SERIES.

4 Quarter Moving Average.

• For calculate the four quarter moving averages and even data series, we should
follow the steps below:

• Step 1: List down the year quarters and observed values as shown in column 1, 2
and 3.

• Step 2: Sum up the first values and divide by 4.Write the answer in between the 2 nd
and 3rd quarter of 2002.

• Step 3: Remove the first value and find the average for the next four values. Write
the answer in between the 3 rd and 4th quarter of 2002 and repeat until the last values
is used. There will be no trend values for the first two and the last two items.

• Step 4: To find the trend, we find the average of the first two values in column 4 and
write the answer in between the two values. This will be the trend values for 3 rd
quarter of 2002. Remove the first value and find the average of the next two values
and write the answer between them and continue until last values of moving average.

85
 Column 1 Column 2 Column 3 Column 4 Column 5

Year Quarter Natural rubber Moving Centered

production (‘ 000) Average Average

2002 I 221 - -

II 183 - -

222.50

III 251 226.75

231.00

IV 235 233.25

235.50

2003 I 255 237.38

239.25

II 201 242.88

246.50

III 266 252.63

258.75

IV 264 266.25

273.75

2004 I 304 280.63

287.5

II 261 283.63

279.75

III 321 - -

IV 233 - -

In order to forecast the trend values, we must first find the average trend increase/decrease
per quarter using the formula below.
Last trend value−First trend value
Trend increase∨decrease=
number of trend −1

283.63−226.75
¿
8−1

= 8.1257
Trend
Forecast trend for first quarter of 2005 = Last trend value + no of step ahead to reach increase/decrease
first quarter of 2005

= 283.63 + (3)8.1257

= 308.0071(‘000)

86
5.6.2 Estimating the Seasonal Variation
Seasonal variations are upwards or downwards swing in the values of data around the trend
line, usually repeated on a regular basis. To calculate seasonal variation, we need to find the
difference between observed data values and trend values.

Example 52

Based on the table below calculate the seasonal index of the number of component for each
week and interpret the result.

Week
Month 1st 2nd 3rd 4th
January - - 420 400
February 350 400 320 280
March 420 400 520 500

Solution:

– The trend values are calculated by using the moving average method.
– The seasonal variations, V considered as percentage of the trend are
calculated using the following formula.

87
Forecasting:

Tabulate the results of seasonal variations. Find the average of these variations and
calculate for seasonal index.

Month 1 2 3 4
Jan
Feb 92.11 113.48 92.42 78.87
March 110.53 92.49
Total 202.64 205.97 92.42 78.87
Average 101.32 102.99 92.42 78.87
Adjustment 1.0650
Seasonal index (%) 107.91 109.68 98.43 84.00

Required total
Adjustment=
Actual total

400
Adjustment= =1 . 0650
375 . 595

Seasonal index = average x adjustment

Interpretation of seasonal index for the first, second, third and fourth quarters are as
follows.

The number of component for the first quarter is 7.91 % above the trend.
The number of components for the second quarter is 9.68 % above the trend.
The number of component for the third quarter is 1.57 % below the trend.
The number of component for the fourth quarter is 16 % below the trend.

In order to forecast;

TxS
Forecast =
100

Example 53

Forecast the number of component for first month of April.

Seasonal Index (for 1st month) = 107.91

Trend for first month of April:

Last trend value−First trend value

Trend increase∨decrease=
number of trend −1

88
 432.5−380
¿
6−1

= 10.5

Trend
Forecast trend for first quarter of 2005 = Last trend value + no of step ahead to reach increase/decrease

first quarter of 2005

= 432.5 + (3)10.5
= 464
464 x 107.91
Forecast the number of component for first month of April =
100
= 500.7024

TUTORIAL 5

1) State which of the four components of time series contribute to the effect of the following
events.
a) The gradual increase in the cost of living in Malaysia from 2004 to 2014.
b) The increase in the demand of masks and gloves during the outbreak of H1N1.
c) The increase in the demand of mandarin oranges before Chinese New Year.
d) The increase in the National Educational Saving Scheme (SSPN) after the Malaysian
government allows a tax relief of not more than RM6000 yearly.
e) The increase in the investment in unit trust when the government allows employees
to withdraw their money from the Employee Provident Fund (EPF) for investment
purposes.

2) Table below relate to units of service demanded of a service provider operating 24

hours a day.

Day Sales (RM’000)

Morning Afternoon Night
1 610 200 490
2 690 220 480
3 750 230 590
4 790 270 570

a) Find the trend values using a three session moving average.

b) Estimate the trend of service demanded for the morning session on 5th day.

3) Table shows the order in RM’000 received by Top Ten Industries in the first 12 weeks of
2015.

Week January February March

1 19 21 23

89
 2 20 22 25
3 22 18 24
4 24 25 30

a) Find the trend of the above data using the method of moving average.
b) Estimate the trend of order received for the 4th week of April.

4) The seasonal indices for expenses of Star Hypermarket from 2013 – 2015 are shown
below.

Season 1 2 3 4
Seasonal Index (RM’000) 102 w 98 z

There are four seasons in a year. If the expenses for the second season are 3%
below the trend, find the values of w and z.

5) The sales, in tons recorded by a company are shown below:

Year Quarter 1 Quarter 2 Quarter 3 Quarter 4

2012 - - 10 16
2013 17 18 13 18
2014 19 21 14 15
2015 18 19 - -

a) Find the trend value using the moving average method.

b) Calculate the seasonal index for each quarter.
c) Forecast the sales for the third and fourth quarter of 2015.

6) Table shows the quarterly electricity account in RM for E & E Co. from 2012 to 2015.

1 2 3 4
2012 - - 1552 1602
2013 1576 1608 1710 1736
2014 1633 1672 1718 1766
2015 1800 1812 - -

a) Find the trend equation using the least squares method.

b) Calculate the seasonal indices for the four quarters.

90
 c) Forecast the sales for the 1st quarter of 2016.

7) The number of four monthly food delivery rider accidents from year 2016 through 2019
are reported at a police station as shown in the table below. Find the value of i), ii), iii),
iv), v), vi), vii) and viii). (Round the answers to 3 decimal places) (Assessment 2 March
21)

Year Term No. of accident Moving average Seasonal variation

2016 Jan - Apr 35 i)
May - Aug 28
Sept - Dec 54 ii) vi)
2017 Jan - Apr 42
May - Aug 37
Sept - Dec 60 iii)
2018 Jan - Apr 49
May - Aug 43 iv) vii)
Sept - Dec 66
2019 Jan - Apr 51
May - Aug 48 v) viii)
Sept - Dec 75

8) The following table represents the seasonal variation of quarterly profit (RM ‘00) for a
construction company. Calculate the seasonal indices for each quarter and explain the
meaning of seasonal index for fourth quarter. (Assessment 2 March 21)

Year Quarter
1 2 3 4
2017 - - 95.840 96.260
2018 102.025 102.183 100.020 98.488
2019 101.371 101.355 100.447 100.197
2020 99.407 100.436 - -

9) The table below shows the number of cans of drink sold over a three- shift period. (Final
Assessment July 21)

Day
Shift
Monday Tuesday Wednesday Thursday Friday
Morning 18 22 9 7 15
Afternoon 19 23 9 8 16
Evening 21 23 10 10 16

i) Find the trend values using the moving average method. Give your answers correct
to 4 decimal places.
ii) Compute the seasonal indices for each shift.

91
 iii) Interpret the seasonal indices for Morning and Evening.
iv) Forecast the sales of cans in Afternoon on Saturday.

10) Determine the appropriate component of time series for the following situations. (Final
Assessment February 21)
a) Increased demand for white shoes before school starts.
b) Decreased number of flight ticket bookings since the Covid19 outbreak.

11) The following data shows the number of COVID-19 patients still isolated at Suria
Hospital as of November 2020. (Final Assessment February 21)

Day
Week
Monday Tuesday Wednesday Thursday Friday
1 12 20 14 50 31
2 11 15 16 24 45
3 14 5 25 28 50

a) Use the above data to find the trend values by using moving average method
(Correct to three decimal places).
b) Find the seasonal index for each day.
c) Interpret the meaning of seasonal index for Wednesday.
d) On which day of the week would you recommend to have the high number of the
e) COVID-19 patients at Suria hospital? Give reason for your answer.
f) Forecast the number of patients who will be seeking treatment on Wednesday in
fourth week of November 2020.

12) Table below is about quarterly production (in ‘000) speedometer brand XPM in factory
from year 2015 to 2019. (Final Assessment July 20)

Year Quarter
1 2 3 4
2015 - 450 408 665
2016 481 401 657 545
2017 467 518 422 607
2018 543 668 614 498
2019 491 431 482 406

a) Plot the actual value on the graph paper.

b) Determine the trend values using moving average method (correct to three decimal
places).
c) Plot the trend value obtained in (b) into graph part (a).
d) Determine the seasonal index.
e) Interpret the fourth quarter of seasonal index.
f) Forecast the first quarter of year 2020 production of speedometer.

13) Define and give ONE (1) example of the following components in time series. (Final
Examination December 19)
a) Trend
b) Seasonal variation

92
14) The following table shows the number of customers registered for RGH swimming
classes from the year 2016 to 2018. (Final Examination December 19)

Year Quarter
1 2 3 4
2016 60 50 40 80
2017 70 40 30 70
2018 80 40 40 90
a) Find the trend values for the number of customers registered for RGH swimming
classes using the moving average method. (Correct to two decimal places).
b) Determine the seasonal index for each quarter.
c) Interpret the seasonal indices obtained for the third and fourth quarter.
d) Forecast the number of customers registered in the third quarter of 2019.

15) The following table shows the number of books sold (in units) by Fury bookstore from
the month of August to October 2018. (Final Examination June 19)

Month Quarter
1 2 3 4
August 230 270 200 250
Septembe 300 310 280 290
r
October 350 300 280 260
a) Find the trend values using the moving average method (correct to two decimal
places)
b) Determine the seasonal index for each week.
c) Interpret the seasonal indices obtained for the second and fourth week.
d) Forecast the number of books sold for the third week of November 2018.

93