UNIT-1

TIME SERIES

1.1. Meaning

An arrangement of statistical data in accordance with time of occurrence or in a

chronological order is called a time series. The numerical data which we get at different
points of time-the set of observations-is known as time series.

In time series analysis, current data in a series may be compared with past data in the
same series. We may also compare the development of two or more series over time.
These comparisons may afford important guide lines for the individual firm. In Economics,
statistics and commerce it plays an important role.

1.2 Definition and Examples

A time series is a set of observations made at specified times and arranged in a

chronological order.

For example, if we observe agricultural production, sales, National Income etc., over
a period of time, say over the last 3 or 5 years, the set of observations is called time series.
Thus a time series is a set of time, quantitative readings of some various recorded at equal
intervals of time. The interval may be an hour, a day, a week, a month, or a calendar year.
Hourly temperature reading, daily sales in a shop, weekly sales in a shop, weekly sales in a
market, monthly production in an industry, yearly agricultural production, population growth
in ten years, are examples of time series.

From the comparison of past data with current data, we may seek to establish what
development may be expected in future. The analysis of time series is done mainly for the
purpose of forecasts and for evaluating the past performances. The chronological variations
will be object of our study in time series analysis.

The essential requirements of a time series are:

 The time gap, between various values must be as far as possible, equal.

2
  It must consist of a homogeneous set of values.
 Data must be available for a long period.

symbolically if „t‟ stands for time and „yt‟ represents the value at time t then the paired values
(t, yt) represents a time series data.

Ex 1: Production of rice in Tamilnadu for the period from2010-11 to 2016-17.

Table 1.1. Production of rice in Tamilnadu (in ‘000 metric tons)

Year Production
2010-11 400
2011-12 450
2012-13 440
2013-14 420
2014-15 460
2016-17 520

1.3 Uses of Time Series

The analysis of time series is of great significance not only to the economists and
business man but also to the scientists, astronomists, geologists, sociologists, biologists,
research worker etc. In the view of following reasons

 It helps in understanding past behavior.

 It helps in planning future operations.
 It helps in evaluating current accomplishments.
 It facilitates comparison.

1.4 Components of Time Series

The values of a time series may be affected by the number of movements or

fluctuations, which are its characteristics. The types of movements characterizing a time
series are called components of time series or elements of a time series.

These are four types

3
  Secular Trend
 Seasonal Variations
 Cyclical Variations
 Irregular Variations

Secular Trend
Secular Trend is also called long term trend or simply trend. The secular trend refers
to the general tendency of data to grow or decline over a long period of time. For example the
population of India over years shows a definite rising tendency. The death rate in the country
after independence shows a falling tendency because of advancement of literacy and medical
facilities. Here long period of time does not mean as several years. Whether a particular
period can be regarded as long period or not in the study of secular trend depends upon the
nature of data. For example if we are studying the figures of sales of cloth store for 1996-
1997 and we find that in 1997 the sales have gone up, this increase cannot be called as
secular trend because it is too short period of time to conclude that the sales are showing the
increasing tendency.

On the other hand, if we put strong germicide into a bacterial culture, and count the
number of organisms still alive after each 10 seconds for 5 minutes, those 30 observations
showing a general pattern would be called secular movement.

Mathematically the secular trend may be classified into two types

1. Linear Trend
2. Curvi-Linear Trend or Non-Linear Trend.

4
 If one plots the trend values for the time series on a graph paper and if it gives a
straight line then it is called a linear trend i.e. in linear trend the rate of change is constant
where as in non-linear trend there is varying rate of change.

Upward linear trend Non-linear trend

Value
of
Value
Variable
of
variable

Down ward linear trend

Time period Time

period Linear trend Nonlinear trend

Figure 1.1. Linear Trend and Nonlinear Trend

Seasonal Variations

Seasonal variations occur in the time series due to the rhythmic forces which occurs in
a regular and a periodic manner with in a period of less than one year. Seasonal variations
occur during a period of one year and have the same pattern year after year. Here the period
of time may be monthly, weekly or hourly. But if the figure is given in yearly terms then
seasonal fluctuations does not exist. There occur seasonal fluctuations in a time series due to
two factors.

 Due to natural forces

 Manmade convention.

The most important factor causing seasonal variations is the climate changes in the
climate and weather conditions such as rain fall, humidity, heat etc. act on different products
and industries differently. For example during winter there is greater demand for woolen
clothes, hot drinks etc. Where as in summer cotton clothes, cold drinks have a greater sale
and in rainy season umbrellas and rain coats have greater demand.

5
 Though nature is primarily responsible for seasonal variation in time series, customs,
traditions and habits also have their impact. For example on occasions like dipawali,
dusserah, Christmas etc. there is a big demand for sweets and clothes etc., there is a large
demand for books and stationary in the first few months of the opening of schools and
colleges.

Cyclical Variations or Oscillatory Variation

This is a short term variation occurs for a period of more than one year. The rhythmic
movements in a time series with a period of oscillation( repeated again and again in same
manner) more than one year is called a cyclical variation and the period is called a cycle. The
time series related to business and economics show some kind of cyclical variations.

One of the best examples for cyclical variations is „Business Cycle‟. In this cycle
there are four well defined periods or phases.

 Boom
 Decline
 Depression
 Improvement

Figure 2. Phases of Business Cycle

Irregular Variation
It is also called Erratic, Accidental or Random Variations. The three variations trend,
seasonal and cyclical variations are called as regular variations, but almost all the time series
including the regular variation contain another variation called as random variation. This type
of fluctuations occurs in random way or irregular ways which are unforeseen, unpredictable
and due to some irregular circumstances which are beyond the control of human being such

6
as earth quakes, wars, floods, famines, lockouts, etc. These factors affect the time series in
the irregular ways. These irregular variations are not so significant like other fluctuations.

1.5 Mathematical Model

In classical analysis, it is assumed that some type of relationship exists among the
four components of time series. Analysis of time series requires decomposition of a series, to
decompose a series we must assume that some type of relationship exists among the four
components contained in it.

The value Yt of a time series at any time t can be expressed as the

combinations of factors that can be attributed to the various components. These combinations
are called as models and these are two types.

 Additive model
 Multiplicative model

Additive model
Yt  Tt  S t  Ct  Rt
In additive model

Where Tt  Trend value at time t

S t  Seasonal component

Ct  Cyclical component

Rt  Irregular component

But if the data is in the yearly form then seasonal variation does not exist, so in that situation
Yt  Tt  Ct  Rt

Generally the cyclical fluctuations have positive or negative value according to

whether it is in above or below the normal phase of cycle.

Multiplicative model:
In multiplicative model Yt  Tt .St .C t .R t .

The multiplicative model can be put in additive model by taking log both sides.
However most business analysis uses the multiplicative model and finds it more appropriate
to analyze business situations.

According to this model, the simple

7
 One of the most important tasks before economists and businessmen these days is to
make estimates for the future. For example, a businessman is interested in finding out his
likely sales in the year 2016 or as a long-term planning in 2025 or the year 2030 so that he
could adjust his production accordingly and avoid the possibility of either unsold stocks or
inadequate production to meet the demand. Similarly, an economist is interested in estimating
the likely population in the coming year so that proper planning can be carried out with
regard to food supply, jobs for the people, etc. However, the first step in making estimates for
the future consists of gathering information from the past. In this connection one usually
deals with statistical data which are collected, observed or recorded at successive intervals of
time. Such data are generally referred to as „time series‟. Thus when we observe numerical
data at different points of time the set of observations is known as time series. For example if
we observe production, sales, population, imports, exports, etc. at different points of time,
say, over the last 5 or 10 years, the set of observations formed shall constitute time series.
Hence, in the analysis of time series, time is the most important factor because the variable is
related to time which may be either year, month, week, day and hour or even- minutes or
seconds.

1.6 Measurement of Secular trend:

Secular trend is a long term movement in a time series. This component represents
basic tendency of the series. The following methods are generally used to determine trend in
any given time series. The following methods are generally used to determine trend in any
given time series.

 Graphic method or eye inspection method

 Semi average method
 Method of moving average
 Method of least squares

Graphic method or eye inspection method

Graphic method is the simplest of all methods and easy to understand. The method is
as follows. First plot the given time series data on a graph. Then a smooth free hand curve is
drawn through the plotted points in such a way that it represents general tendency of the
series. As the curve is drawn through eye inspection, this is also called as eye-inspection
method. The graphic method removes the short term variations to show the basic tendency of
the data. The trend line drawn through the graphic method can be extended further to predict

8
or estimate values for the future time periods. As the method is subjective the prediction may
not be reliable.

1400

1200
production of cotton

1000

800
Series2
600

400

200

0
1971 1972 1973 1974 1975 1976 1977 1978
year

Figure 1.3. Graphic method for thee production of cotton base on year

Advantages
 It is very simplest method for study trend values and easy to draw trend.
 Sometimes the trend line drawn by the statistician experienced in computing trend
may be considered better than a trend line fitted by the use of a mathematical
formula.
 Although the free hand curves method is not recommended for beginners, it has
considerable merits in the hands of experienced statisticians and widely used in
applied situations.

Disadvantages:
 This method is highly subjective and curve varies from person to person who
draws it.
 The work must be handled by skilled and experienced people.
 Since the method is subjective, the prediction may not be reliable.
 While drawing a trend line through this method a careful job has to be done.

Method of Semi Averages:

In this method the whole data is divided in two equal parts with respect to time.
For example if we are given data from 1999 to 2016 i.e. over a period of 18 years the two
equal parts will be first nine years i.e. from 1999 to 2007 and 2008 to 2016. In case of odd

9
number of years like 9, 13, 17 etc. two equal parts can be made simply by omitting the
middle year. For example if the data are given for 19 years from 1998 to 2016 the two equal
parts would be from 1998 to 2006 and from 2008 to 2016, the middle year 2007 will be
omitted. After the data have been divided into two parts, an average (arithmetic mean) of
each part is obtained. We thus get two points. Each point is plotted against the mid year of the
each part. Then these two points are joined by a straight line which gives us the trend line.
The line can be extended downwards or upwards to get intermediate values or to predict
future values.

Example:
Year Production Semi averages

2001 40
2002 45 40  45  40  42
 41.75
2003 40 4

2004 42
2005 46
2006 52 46  52  56  61
 53.75
2007 56 4

2008 61

Thus we get two points 41.75 and 53.75 which shall be plotted corresponding to their
middle years i.e. 2002.5 and 2006.5. By joining these points we shall obtain the required
trend line. This line can be extended and can be used either for prediction or for determining
intermediate values.

Advantages:

 This method is simple to understand as compare to moving average method

and method of least squares.
 This is an objective method of measuring trend as everyone who applies this
method is bound to get the same result.

Disadvantages:

 The method assumes straight line relationship between the plotted points
regardless of the fact whether that relationship exists or not.
10
  The main drawback of this method is if we add some more data to the original
data then whole calculation is to be done again for the new data to get the
trend values and the trend line also changes.
 As the Arithmetic Mean of each half is calculated, an extreme value in any
half will greatly affect the points and hence trend calculated through these
points may not be precise enough for forecasting the future.

Method of Moving Average:

It is a method for computing trend values in a time series which eliminates the short
term and random fluctuations from the time series by means of moving average. Moving
average of a period m is a series of successive arithmetic means of m terms at a time starting
with 1st, 2nd , 3rd and so on. The first average is the mean of first m terms; the second average
is the mean of 2nd term to (m+1)th term and 3rd average is the mean of 3rd term to (m+2)th
term and so on.

If m is odd then the moving average is placed against the mid value of the time
interval it covers. But if m is even then the moving average lies between the two middle
periods which does not correspond to any time period. So further steps has to be taken to
place the moving average to a particular period of time. For that we take 2-yearly moving
average of the moving averages which correspond to a particular time period. The resultant
moving averages are the trend values.

Ex:1 Calculate 3-yearly moving average for the following data.

Years Production 3-yearly moving avg (trend values)
2001-02 40

2002-03 45 (40+45+40)/3 = 41.67

2003-04 40 (45+40+42)/3 = 42.33

2004-05 42 (40+42+46)/3 = 42.67

2005-06 46 (42+46+52)/3 = 46.67

2006-07 52 (46+52+56)/3 = 51.33

2007-08 56 (52+56+61)/3 = 56.33

2008-09 61

11
Ex :2 Calculate 4-yearly moving average for the following data.

Years Production 4-yearly moving avg 2-yealry moving avg

(trend values)
2001-02 40
2002-03 45
(40+45+40+42)/3 = 41.75
2003-04 40 42.5
(45+40+42+46)/3 = 43.15
2004-05 42 44.12
(40+42+46+52)/3 = 45
2005-06 46 47
(42+46+52+56)/3 = 49
2006-07 52 51.38
(46+52+56+61)/3 = 53.75
2007-08 56
2008-09 61

Advantages:

 This method is simple to under stand and easy to execute.

 It has the flexibility in application in the sense that if we add data for a few more time
periods to the original data, the previous calculations are not affected and we get a
few more trend values.
 It gives a correct picture of the long term trend if the trend is linear.
 If the period of moving average coincides with the period of oscillation (cycle), the
periodic fluctuations are eliminated.
 The moving average has the advantage that it follows the general movements of the
data and that its shape is determined by the data rather than the statistician‟s choice of
mathematical function.

Disadvantages:
 For a moving average of 2m+1, one does not get trend values for first m and last m
periods.

12
  As the trend path does not correspond to any mathematical; function, it cannot be
used for forecasting or predicting values for future periods.
 If the trend is not linear, the trend values calculated through moving averages may not
show the true tendency of data.
 The choice of the period is sometimes left to the human judgment and hence may
carry the effect of human bias.

Method of Least Squares:

This method is most widely used in practice. It is mathematical method and with its
help a trend line is fitted to the data in such a manner that the following two conditions are
satisfied.

1.  (Y  Y )  0 i.e.
c the sum of the deviations of the actual values of Y and the

computed values of Y is zero.

 (Y  Y )
2
2. c is least, i.e. the sum of the squares of the deviations of the actual values

and the computed values is least.

The line obtained by this method is called as the “line of best fit”.

This method of least squares may be used either to fit a straight line trend or a parabolic
trend.

Fitting of a straight line trend by the method of least squares:

Let Yt be the value of the time series at time t. Thus Yt is the independent variable
depending on t.

Assume a straight line trend to be of the form Ytc  a  bt …………. (1)

Where Ytc is used to designate the trend values to distinguish from the actual Yt values, a is

the Y-intercept and b is the slope of the trend line.

Now the values of a and b to be estimated from the given time series data by the method of
least squares.

In this method we have to find out a and b values such that the sum of the squares of
Yt Ytc
the deviations of the actual values and the computed values is least.

13
 i.e. S   (Yt  Ytc ) should be least
2

i.e. S   (Yt  a  bt) ………… (2) Should be least

2

Now differentiating partially (2) w.r.to a and equating to zero we get

S
 2 (Yt  a  bt )(1)  0
a

  (Yt  a  bt )  0
  Yt   a  b t

 Yt  na  bt
……………….. (3)

Now differentiating partially (2) w.r.to b and equating to zero we get

S
 2 (Yt  a  bt )(t )  0
b

  t (Yt  a  bt )  0
  tYt  a  t  b t 2
……………….. (4)

The equations (3) and (4) are called „normal equations‟

Solving these two equations we get the values of a and b say â and b̂ .

Now putting these two values in the equation (1) we get

Ytc  aˆ  bˆt

which is the required straight line trend equation.

Note: The method for assessing the appropriateness of the straight line modal is the method
of first differences. If the differences between successive observations of a series are constant
(nearly constant) the straight line should be taken to be an appropriate representation of the
trend component.

14
To find a and b we have two normal equations

Y  na  b X
 XY  a X  b X 2

15
To find a and b we have two normal equations

Y  na  b X
 XY  a X  b X 2

The required line equation is

The trend values for various years are

16
and so on

Fitting of a parabolic trend by the method of least squares

Let Yt be the value of the time series at time t. Thus Yt is the independent variable
depending on t.

Assume a parabolic trend to be of the form Ytc  a  bt  ct 2 …………. (1)

Now the values of a, b and c to be estimated from the given time series data by the
method of least squares.

In this method we have to find out a, b and c values such that the sum of the squares

of the deviations of the actual values Yt and the computed values Ytc is least.

i.e. S   (Yt  Ytc ) should be least

2

i.e. S   (Yt  a  bt ) ………… (2) Should be least

2

Now differentiating partially (2) w.r.to a and equating to zero we get

S
 2 (Yt  a  bt  ct 2 )(1)  0
a

17
   (Yt  a  bt  ct 2 )  0

  Yt   a  b t  c  t 2

 Yt  na  bt  ct 2

……………….. (3)

Now differentiating partially (2) w.r.to b and equating to zero we get

S
 2 (Yt  a  bt  ct 2 )(t )  0
b

  t (Yt  a  bt  ct 2 )  0
……………….. (4)
  tYt  a  t  b t  c  t
2 3

Now differentiating partially (2) w.r.to c and equating to zero we get

S
 2 (Yt  a  bt  ct 2 )(t 2 )  0
c

  t 2 (Yt  a  bt  ct 2 )  0
  t 2Yt  a  t 2  b t 3  c  t 4
……………….. (5)

The equations (3), (4) and (5) are called „normal equations‟

Solving these three equations we get the values of a, b and c say aˆ , bˆ and ĉ .

Now putting these three values in the equation (1) we get

Ytc  aˆ  bˆt  cˆt 2

Which is the required parabolic trend equation

Note: The method for assessing the appropriateness of the second degree equation is
the method of second differences. If the differences are taken of the first differences and the
results are constant (nearly constant) the second degree equation be taken to be an appropriate
representation of the trend component.

18
19
Advantages

 This is a mathematical method of measuring trend and as such there is no

possibility of subjectiveness i.e. everyone who uses this method will get same
trend line.
 The line obtained by this method is called the line of best fit.
 Trend values can be obtained for all the given time periods in the series.

Disadvantages

 Great care should be exercised in selecting the type of trend curve to be fitted
i.e. linear, parabolic or some other type. Carelessness in this respect may lead
to wrong results.
 The method is more tedious and time consuming.

20
  Predictions are based on only long term variations i.e trend and the impact of
cyclical, seasonal and irregular variations is ignored.
 This method can not be used to fit the growth curves like Gompertz
 1 

curve Y  K a
bX
 Y 
, logistic curve 
X 
K  ab  etc.

Question bank:
1) Define a time series. Discuss its main components.
2) Define secular trend of a time series and explain methods that are used in isolating it.
3) Explain the method of moving average for the determination of trend. What are the
advantages and disadvantages of this method?

4) What are the advantages and disadvantages of the graphic method and least square
method in trend analysis?

5) Explain briefly the method of moving averages for calculating the trend.
6) How does analysis of time series help business forecasting?
7) Distinguish between secular trend, seasonal variations and cyclical fluctuations. Discuss
various methods of measuring each.

8) Explain briefly the additive and multiplicative models of time series. Which of these
models is more popular in practice and why?

9) Explain how you would determine seasonal variation by 12-monthly moving average.
10) What are the various methods for determining trend in a time series?
11) Describe in detail the method of least squares for determining trend.
12) The production data on steel in a factory in the past 10 years are given below:
Year : 1987 1988 1989 1990 1991 1992 1993 1994 1995
1996

Production : 75 86 98 90 96 108 124 140 150

165
(tonnes)

Fit a staright line trend and tabulate the trend values. What is the expected production in
the year 1997 on the basis of trend? (Y  113.2  5.08 X ; Y  169.98)

13) Fit a staright line trend of the following data by least square method. Also find an
estimate for the year 1997;

21
 Year : 1989 1990 1991 1992 1993 1994 1995 1996

Production : 12 13 13 16 19 23 21 23
(tonnes) (Y  17.5  .893 X ; Y1997  25.54)

14) Fit a staright line trend by the method of least square to the following data. Also find an
estimate for the year 2000;

Year : 1990 1991 1992 1993 1994 1995 1996 1997

Production : 38 40 65 72 69 67 95 104
(tonnes) (Y  68.75  4.404 X ; )

15) From the following data, calculate trend by 4 yearly moving average and find short-term
oscillations:

Year Production(tonnes) Year

Production(tonnes)
1984 5 1990 9
1985 6 1991 10
1986 7 1992 9
1987 7 1993 10
1988 6 1994 11
1989 8 1995 11

22