1

Chapter 8
TIME SERIES ANALYSIS
Time series is a set of data for a particular period of time. A time series
is a set of statistics, usually collected at regular intervals. The analysis of
time series helps in forecasting. Forecasting is the process of predicting a
future event. Time series analysis is a method of analyzing the time series
data for the purpose of understanding the characteristics of the data.
Further, it helps in predicting or projecting future values based on the
observed historical data. Let us look into the example given below:

The chart shows the population growth in India for a period of time from
2000 to 2050. Hence, from time series analysis we can understand the
present data and we can predict the future trend also.
 2

Utility of time series:

1. It helps in understanding past behavior of data.
2. It helps in planning future operations. For eg:- sales for past 3 months
are down trend or falls down hence, the marketing campaign is to be
strengthened.
3. It helps in evaluating current accomplishments.
4. It facilitates comparison of data across different period of time.
Components of time series:
1. Secular trend (common or routine trend)
It is persistent movements of data and overall upward or
downward pattern is called secular trend and it is due to population,
technology etc. It is a long term growth or decline of a series of data. It
will take several years duration.
2. Seasonal trend (changes due to season)
The seasonal variation of a time series is a pattern of change
that recurs regularly over time and the pattern starts repeating itself at
a fixed time each year. It is a regular pattern of up & down fluctuations
due to weather, customs etc. and it occurs within one year. For
example, the demand for air coolers will be higher during summer and
heaters during winter seasons.

3. Cyclical trend (eg. Change in the data due to economic cycle)

The data are featured by repeating up & down movements due
to interactions of factors influencing economy. It occurs usually 2-10
years duration.
4. Irregular trend (reasons for changes may be unexpected)
These are the fluctuations of data due to random or unforeseen
reasons and they are erratic and unsystematic fluctuations due to the
reasons like strike, natural calamities etc. They are occurring for a very
short duration and non repetitive in nature.
 3

Graphical representation of Components of time series behavior:

a) Basic level data

b) Trend Movement

c) Cyclic Fluctuations of data

 4

Measurement of trend (Methods of measurements)

1. Freehand method / graphic method:
Freehand drawing is a simple method of measuring secular trend
in which the actual data are plotted on a graph and joining the points
plotted successively or one after the other starting from first to last with a
straight line. This line is called actual data line. Then, one can draw the
trend line by a dotted line in such a way that it should be drawn on the
actual data line and the trend line should pass through all the segments
of the actual data line. And by projecting the trend line one can predict
the value for future. The process is somewhat subjective and does not
require any mathematical calculation.

Example : Fit the trend line by the freehand method

Year production of steel in billion tones
1995 20
1996 22
1997 24
1998 21
1999 23
2000 25
2001 23
2002 26
2003 25
2004 30
 5

35
production of steel in billion tonns 30
25
20
15
10 Production
Linear (Production)
5 Exponential (Produc-
tion)
0
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year

2. Semi – average method:

The method of semi average is the easiest of all the scientific
methods of describing secular trend. In this method, the entire data are
divided into two equal halves and the average of these two groups is
found out. Then, the two averages are plotted on the graph with respect
to the middle year of the each group. The trend line is drawn taking these
averages as the base and the trend will pass through these average
values and also on the actual data line. And by extending the trend line
one can predict the value for future.
 6

The process will be more understandable by the following example.

Example:
Draw the trend line by semi-average method:
Year production of steel in billion tones
1995 20
1996 22
1997 24 22
1998 21
1999 23
2000 25
2001 23
2002 26 25.8
2003 25
2004 30

20+22+24 +21+ ¿ ¿ = 22
25+23+26+ 25
= 25
4 4
35
30
25
20 Actual Data Line
15 Semi Averages
Exponential (Semi Averages)
10
5
0
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

3. Moving average method:

The method of moving average helps to get a smoother trend line
than that of other methods. The trend values obtained through moving
average method will lower the fluctuations of trend data so that the trend
 7

value will be closer to the actual data. In this method, first of all, one has
to decide the period for calculating average values like three yearly or
five yearly or four yearly period. The length of the period may be decided
according to the business cycle. After deciding the period, for each period
the average value is calculated by averaging the actual data of each
period and the average is written against the middle year of the every
period. For example, in case of three yearly moving average the trend
values are calculated as given below:
X 1+ X 2+ X 3 X 4+ X 5+ X 6 X 7+ X 8+ X 9
, and so on.
3 3 3
For four yearly moving average the calculation is as follows:

X 1+ X 2+ X 3+ X 4 X 5+ X 6+ X 7+ X 8 X 9+ X 10+ X 11+ X 12
, ,
4 4 4
X1, X2,X3….. are the given time series data for every consecutive years.

Example
Calculate three yearly and five yearly moving averages for the sales data
given below and draw the trend line by moving average method:

Year Sales (in ’00,000)

2000 12
2001 17
2002 20
2003 21
2004 26
2005 25
2006 28
2007 31
2008 30
2009 35
2010 40
 8

Solution
Calculation of three yearly and five yearly moving averages:
Year Sales 3 yearly 3 yearly 5 yearly 5 yearly
(in’00,000) total average total
average
2000 12 -- -- -- --
2001 17 49 16.33 -- --
2002 20 58 19.33 96
19.2
2003 21 67 22.33 109
21.8
2004 26 72 24 120 24
2005 25 79 26.33 131
26.2
2006 28 84 28 140 28
2007 31 89 29.67 149
29.8
2008 30 96 32 164 32.8
2009 35 105 35 -- --
2010 40 -- -- -- --

4. Method of Least Squares

This is a mathematical method in which a straight line trend is
drawn with the equation Yc = a + bX.
Equations
Y = a+bX
∑y ∑ XY
a = , b = 2
N ∑x
These formulae of a and b are derived from solving the following two
normal equations.
∑y = Na + b ∑X
∑xy = a∑x + b∑ x 2
 9

Example:
Below are the figures of production (in thousand Quintals) of sugar
Year Production (IN thousand
Quintaals)
1997 80
1998 90
1999 92
2000 83
2001 94
2002 99
2003 92
1. Fit a straight line trend to these figures
2. Plot these figures on a graph & show the trend line
3. Find out production for the year 2005
Solution
year Y X Xy x
2
Yc (trend)
97 80 -3 -240 9 84
98 90 -2 -180 4 86
99 92 -1 -92 1 88
2000 83 0 0 0 90
2001 94 1 94 1 92
2002 99 2 198 4 94
2003 92 3 276 9 96
2004 630 4 56 28
2005 5 100

∑Y 630
a = =
N 7

= 90
 10

∑ xy
b = 2
∑x

56
=
28
= 2
Y = a+bx
Y = 90+2x
Take x = -3
Y = 90+2x-3
= 90-6
= 84
The production for the year 2005 = 100 thousand quintals
2. The sale of a company in lakhs rupees are given below. Find the trend
value by least square method and estimate the sales value for 2016
and 2018.

Year Sales
2008 34
2009 52
2010 68
2011 105
2012 140
2013 180
2014 284
 11

Answer
Year Y X xy x
2
yc
2008 34 -3 -102 9 7.79
2009 52 -2 -104 4 46.29
2010 68 -1 -68 1 84.79
2011 105 0 0 0 123.29
2012 140 1 140 1 161.79
2013 180 2 360 4 200.29
2014 284 3 852 9 238.79
2016 863 5 1078 28 315.79
2018 7 312.79

∑Y
a =
N
863
=
7
= 123.29
∑ xy
b = 2
∑x
1078
=
28
= 38.5
Y = a + bx
= 123.29+38.5x
The sale value for 2016 = Rs. 315.79 Lakhs rupees
The sale value for 2018 =Rs. 392.79 Lakhs rupees