Statistics for Management Unit 14

Unit 14 Time Series Analysis

Structure:
14.1 Introduction
Objectives
Relevance
14.2 Time Series Analysis
14.3 Utility of the Time Series
14.4 Components of Time Series
Long term trend or secular trend
Seasonal variations
Cyclic variations
Random variations
14.5 Methods of Measuring Trend
Free hand or graphic method
Semi averages method
Method of moving averages
Method of least squares
14.6 Mathematical Models for Time Series
Additive model
Multiplicative model
14.7 Editing of Time Series
14.8 Measurement of Seasonal Variation
Simple average method
Ratio to moving averages method
Chain or link relative method
Ratio to trend method
14.9 Forecasting Methods Using Time Series
Mean forecast
Naive forecast
Linear trend forecast
Non-linear trend forecast
Forecasting with exponential smoothing
14.10 Summary
14.11 Glossary
14.12 Terminal Questions
14.13 Answers
14.14 Case Study
Manipal University Jaipur Page No. 495
Statistics for Management Unit 14

14.1 Introduction
In the previous unit ‘Business Forecasting’, you have studied about the
ways of forecasting business events successfully. You also studied about
the different methods available for forecasting. In this unit, you will study
about the time series analysis and different components of time series. You
will also study about the forecasting methods using time series.
A time series is a set of numerical values of a given variable listed at
successive intervals of time, which means that, data regarding the variable
is listed in chronological order. Usually, the interval of time is taken as
uniform.
Yearly production of wheat in the country, hourly temperature of a city,
bimonthly electricity bills are all examples of time series. Almost all data like
industrial production, agricultural production, exports, imports, dairy
products can be arranged in chronological order.
Objectives:
After studying this unit, you should be able to
 analyse the time series
 describe different components of time series
 describe the forecasting methods
 apply time series analysis in business scenarios
14.1.1 Relevance
The 1990s brought a heightened awareness of an increased concern over
pollution in various forms in the United States. Air pollution is one of the
main areas of environmental concern. The U.S Environmental Protection
Agency (EPA) monitors the quality of air around the country. Some of the air
pollutants monitored include carbon monoxide emissions, nitrogen oxide
emissions, volatile organic compounds, sulphur dioxide emissions, etc. The
substances in these pollutants cause cancer and respiratory problems. If the
data is given for 15 years of period (1985-1999), then the question is to find
whether the air quality in U.S has been improving or deteriorating over time.
Managerial and statistical questions:
1. Is it possible to forecast the emissions of carbon monoxide or nitrogen
oxides for the year 2004-2007, or 2020 using the available data?
2. What techniques best forecast the emissions of carbon monoxide or
nitrogen oxides in the future?
Manipal University Jaipur Page No. 496
Statistics for Management Unit 14

These questions are best answered by time series analysis.

14.2 Time Series Analysis

Given a time series, we need to study about the forces that influence the
variations in time series and the behaviour of phenomenon over the given
period of time. For example, consider the sales of T.V sets (in thousands) by
a producing company. The table 14.1 depicts the sales data of TV sets sold
from 1995 to 2000.
Table 14.1: Sales Data of TV Sets Sold From 1995 to 2000
Year 1995 1996 1997 1998 1999 2000
Number of TV sets
12 14 16 12 10 18
sold (in thousands)

Let us analyse the above data and give some trends regarding the sales.
For example, the company would like to know why sales dropped in 1998
and 1999 and why did it increase. In other words, the company would like to
analyse the various forces that affect the sales.
There can be changes in the values of the variable recorder over different
points of time due to various forces. Analysing the effect of all such forces
on the values of the variable is generally known as the analysis of time
series. Broadly, the following are the four types of changes in the values of
the variable:
i) Changes which generally occur due to general tendency of the data to
increase or decrease
ii) Changes which occur due to change in climate, weather conditions
and festivals
iii) Changes which occur due to booms and depressions
iv) Changes which occur due to some unpredictable forces like floods,
famines and earthquakes

14.3 Utility of the Time Series

The following are the possible uses of the time series:
i. The comparative study of behaviour of the variable over different
periods of time can be done. The variable may be export figures,
quantity of industrial production etc.

Manipal University Jaipur Page No. 497

Statistics for Management Unit 14

ii. Forecasting can be done using the time series. By studying the
variations and other behaviour of the variables over a sufficiently long
period of time, it may be possible to forecast the future behaviour of the
variables. However, such a forecast has meaning only if the period of
forecast is a normal period. For example, various five-year plans by the
government of India are formulated by studying the time series and
forecasting.
iii. Study of the time series helps in analysing the post behaviour of the
variables. This helps in identifying the various forces that affect its
behaviour.

14.4 Components of Time Series

The behaviour of a time series over periods of time is called the movement
of the time series. The time series is classified into the following four
components:
i) Long term trend or secular trend
ii) Seasonal variations
iii) Cyclic variations
iv) Random variations
i) Long term trend or secular trend
This refers to the smooth or regular long term growth or decline of the
series. This movement can be characterised by a trend curve. If this curve is
a straight line, then it is called a trend line. If the variable increases over a
long period of time, then it is called an upward trend. If the variable
decreases over a long period of time, then it is called a downward trend. If
the variable moves upward or downward along a straight line then the trend
is called a linear trend, otherwise it is called a non-linear trend.

ii) Seasonal variations

Variations in a time series that are periodic in nature and occur regularly
over short periods of time during a year are called seasonal variations.
These variations are precise and can be forecasted.
The following are examples of seasonal variations in a time series.
i. The prices of vegetables drop down after rainy season or in winter
months and they go up during summer, every year.

Manipal University Jaipur Page No. 498

Statistics for Management Unit 14

ii. The prices of cooking oils reduce after the harvesting of oil seeds and
go up after some time.

iii) Cyclic variations

The long-term oscillations that represent consistent rise and decline in the
values of the variable are called cyclic variations. Since these are long-term
oscillations in the time series, the period of oscillation is usually greater than
one year. The oscillations are either a trend curve or a trend line. The period
of one cycle is the time-distance between two successive peaks or two
successive troughs.
iv) Random variations
Random variations are called irregular movements. Movements that occur
usually in brief periods of time, without any pattern and which are
unpredictable in nature are called irregular movements. These movements
do not have any regular period or time of occurrences. For example, the
effect of national strikes, floods, earthquakes, etc. It is very difficult to study
the behaviour of such a time series.

14.5 Methods of Measuring Trend

We will study the following methods of measuring the trend of a time series:
i. Free hand or graphic methods
ii. Semi averages method
iii. Moving average method
iv. Method of least squares
14.5.1 Free hand or graphic method
This is the simplest method of drawing a trend curve. We plot the values of
the variable against time on a graph paper and join these points. The trend
line is then fitted by inspecting the graph of the time series. Fitting a trend
line by this method is arbitrary. The trend line is drawn such that the
numbers of fluctuations on either side are approximately the same. The
trend line should be a smooth curve.
The free hand method has the following disadvantages:
i. It depends on individual judgement.
ii. It cannot be used for any predictions of trends as drawing the trend
curve is arbitrary

Manipal University Jaipur Page No. 499

Statistics for Management Unit 14

Solved problem 1
Find the trend with the help of free hand curve method for the data depicted
in table 14.2
Table 14.2: Production Data from 1991 to 2001
Year Production Data (in Lakh ton)
1991 15
1992 18
1993 16
1994 22
1995 19
1996 24
1997 20
1998 28
1999 22
2000 30
2001 26

Solution: Figure 14.1 depicts free hand curve of the production data versus
the time period. In the graph, we have taken production data values on
Y-axis and values of time on X-axis.

Fig. 14.1: Free Hand Curve for Solved Problem 1

Manipal University Jaipur Page No. 500

Statistics for Management Unit 14

14.5.2 Semi Averages method

The methods of fitting a linear trend with the help of semi average method
are as follows:
i. When the number of years is even, then the data of the time series is
divided into two equal parts. The total in each of the part is calculated
and then divided by the number of items to obtain arithmetic means of
the two parts. Each average is then centred in the period of time from
which it has been computed and plotted on the graph paper. A straight
line is drawn passing through these points. This is the required trend
line.
ii. When the number of years is odd, then the value of the middle year is
omitted to divide the time series into two equal parts. Then the
preceding procedure is followed.
Solved Problem 2
The figures given below show the export of sugar from India for the years
1971 to 1980. Determine the trend of the following by using semi averages
method.
Table 14.3: Export of Sugar from India from 1971 to 1980
Years Exports (in lakh tonnes)
1971 3.9
1972 1.3
1973 1.1
1974 4.4
1975 9.4
1976 9.6
1977 3.4
1978 2.5
1979 8.6
1980 2.9

The trend of the export of sugar from India can be found by the semi
averages method as follows:
Here the number of years is 10 i.e., it is even. The series is divided into two
halves consisting of the first five years and last five years.

Manipal University Jaipur Page No. 501

Statistics for Management Unit 14

The average of the first five years=1/5 (3.9+1.3+1.1+4.4+9.4)

= 4.02
The average of the second five years=1/5 (9.6+3.4+2.5+8.6+2.9)
= 5.4
Plot the data on a graph paper and join them by dotted lines as shown in the
previous example. Plot the average value of the first half series against the
year 1973 and the average of the last five years against 1978 on the same
graph. On joining these two points, we get the trend line.
The merits and demerits of semi averages method are depicted in
table 14.4.
Table 14.4: Merits and Demerits of Semi averages Method
Merits Demerits

The semi average method is The semi average method assumes a straight
simple. line relationship between the plotted points,
regardless of the fact whether such
relationship exists or not.
The trend line can be extended This method has an in built limitation of
on either side in order to obtain arithmetic mean. This method is not suitable in
past or future estimates. case of very low or very large extreme values.

This is an objective method, as There is no assurance that the influence of

any one applying this method cycle is eliminated.
get the same trend line.

14.5.3 Method of Moving Averages

Moving averages method is used for smoothing the time series. It
smoothens the fluctuations of the data.
When period of Moving Average is odd:
The procedure to determine the trend by this method is depicted in
figure 14.2.
By plotting these trend values (if desired) you can obtain the trend curve,
with the help of which, you can determine the increasing or decreasing
trend. If needed, you can also compute short-term fluctuations by
subtracting the trend values from the actual values.

Manipal University Jaipur Page No. 502

Statistics for Management Unit 14

Fig. 14.2: Procedure for Determining the Trend when Moving Average is Odd

Solved Problem 3
Calculate the 3 yearly Moving Averages of the data depicted in table 14.5.
Table 14.5: Production Data from 1988 to 1996
Year 1988 1989 1990 1991 1992 1993 1994 1995 1996
Production
21 22 23 25 24 22 25 27 26
(in Lakh ton)

Manipal University Jaipur Page No. 503

Statistics for Management Unit 14

Solution: Table 14.5a depicts the calculated values of 3 yearly averages.

Table 14.5a: Calculated Values of 3 Yearly Moving averages

Production 3 –yearly
(Thousand 3 –yearly moving Short term
Year Tonnes) moving average fluctuations
totals (Y - Yc)
Y Yc
1988 21 - - -
1989 22 66 22.00 0
1990 23 70 23.33 - 0.33
1991 25 72 24.00 1.00
1992 24 71 23.67 0.33
1993 22 71 23.67 - 1.67
1994 25 74 24.67 0.33
1995 27 78 26 1.00
1996 26 - - -

When period of moving averages is even:

When period of moving averages is even (such as four years), we compute
the moving averages by using the steps depicted in figure 14.3.

Manipal University Jaipur Page No. 504

Statistics for Management Unit 14

Fig. 14.3: Procedure for Determining the Trend when Moving Average is Even

Manipal University Jaipur Page No. 505

Statistics for Management Unit 14

Table 14.6 depicts the merits and demerits of the moving averages method.

Table 14.6: Merits and Demerits of Moving Averages Method

Merits Demerits
This is a simple method. No functional relationship between the
values and time. Thus, this method is not
helpful in forecasting and predicting the
values on the basis of time.
This method is objective in the No trend values for some years in the
sense that anybody working on a beginning and some in the end.
problem with this method will get
the same results.
This method is used for determining In case of non–linear trend, the values
seasonal, cyclic and irregular obtained by this method are biased in one
variations besides the trend values. or the other direction.
This method is flexible enough to The period selection of moving average is
add more figures to the data a difficult task. Hence, great care has to be
because the entire calculations are taken in period selection, particularly when
not changed. there is no business cycle during that time.
If the period of moving averages
coincides with the period of cyclic
fluctuations in the data, such
fluctuations are automatically
eliminated.

Solved problem 4
The following table gives the average monthly production
(in thousands) of new passenger cars in the period 1976-1985. Calculate
four yearly moving averages.
Table 14.6: Average Monthly Production of Cars
Year Average monthly production of new cars
1976 708
1977 767
1978 764
1979 702
1980 533
1981 521
1982 421
1983 562
1984 635
1985 667
Manipal University Jaipur Page No. 506
Statistics for Management Unit 14

Solution:
The following table depicts four yearly moving averages.
Table 14.6a: Four yearly moving averages

(1) (2) (3) (4) (5) (6)

Year Average 4 yearly 4 yearly 2 item Centered
monthly moving moving moving 4 yearly
production Total average total moving
of new (3) / 4 average
cars (5) / 2
1976 708

1977 767
2941 735.25
1978 764 713.375
2766 691.5 1426.75
1979 702 660.75
2520 630 1321.5
1980 533 587.125
1174.25
2177 544.25
1981 521 1053.5 526.75
2037 509.25
1982 421 1044 522
2139 534.75
1983 562 1106 553
2285 571.25
1984 635

1985 667

14.5.4 Method of least squares

In this method, the trend curve is determined by fitting a mathematical
equation. This method is more accurate and precise and can be used even
for forecasting. We can fit either a straight line or a parabolic curve from the
given data by this method.

Key Statistic
Let ‘Y’ be the actual values of ‘Y’ and ‘Yc’ be the computed values of ‘Y’ for
a given value of ‘X’.
Let ‘Y = a + bX’ be a straight line to be fitted for trend. To find the values of
‘a’ and ‘b’, such that the sum of squares of differences of the actual and
computed values of ‘Y’ is least, that is,

 Y  Y 
2
c is least

Manipal University Jaipur Page No. 507

Statistics for Management Unit 14

where, the condition

 Y  Y   0 is satisfied,
c

is known as method of least squares. The line obtained by the method is

known as the ‘line of best fit.’
For a given time series data, to find a linear trend, the values of ‘a’ and ‘b’
are obtained by the normal equations.
  a b  
  a  b  2
where, N is the number of pairs for which data are given. Here ‘a’ is
intercept of the line on the y – axis and ‘b’ is the slope of the line. ‘b’ is also
known as growth rate (if b > 0) or decline rate (if b< 0), ‘b’ gives the change
in the value of ‘Y’, for per unit change in the value of ‘X’.
1. Direct method
The procedure to be followed is described below.
i) Convert the years into natural numbers (1, 2, 3……) and denote by ‘X’
and find X.
ii) Find the squares of ‘X’ values and obtain X2.
iii) Multiply the X – values with corresponding Y – values and obtain XY
iv) Add the values of Y to obtain Y.
v) Put these values in the two normal equations and solve for ‘a’ and ‘b’.
vi) Substitute these values of ‘a’ and ‘b’ in ‘Y = a + bX’ and then find trend
values for various values of ‘X’.
2. Short cut method
The calculations are simplified when the mid-point in time is taken as origin
so that:  X = 0

When,  X = 0 then normal equations reduce to:

Y
 Y  a , therefore, a
N

XY
 XY  b X 2
, therefore, b 
X 2

Manipal University Jaipur Page No. 508

Statistics for Management Unit 14

Solved problem 5
The production of pig iron and ferro alloys in thousand metric tons in India is
as given below. Find the trend line by the method of least squares.
Table 14.7: Production data
Year Production
1976 672
1977 824
1978 967
1979 1204
1980 1464
1981 1758
1982 2057
Solution:
The trend line can be fitted by using the method of least squares for the
given data.
Table 14.7a: Calculation for trend line

Production
Year X= Year - 1979 XY X2
Y
1976 672 -3 -2016 9
1977 824 -2 -1648 4
1978 967 -1 -967 1
1979 1204 0 0 0
1980 1464 1 1464 1
1981 1758 2 3516 4
1982 2057 3 6171 9
Total ∑Y= 8946 ∑X=0 ∑XY= 6520 ∑X =28
2

When  X = 0, then normal equations reduce to:

Y 8946
 Y  a , therefore, a
N
=
7
= 1278

XY 6520
 XY  b X 2
, therefore, b 
X 2

28
 232 .9

The estimated equation of the trend line is given by

Y  a  bX  1278  232 .9 X

Manipal University Jaipur Page No. 509

Statistics for Management Unit 14

The merits and demerits of method of least squares are displayed in

the table 14.8.
Table 14.8: Merits and Demerits of Direct Method of Least Squares

Merits Demerits
This method is a completely It requires many calculations and is tedious
objective method. and complicated.
This method gives the trend If even a single item is added to the series
values for the entire time period. a new equation has to be formed.
This method can be used to Future forecasts made by this method are
forecast future trend because based only on trend values. Seasonal,
trend line establishes a functional cyclical or irregular variations are ignored.
relationship between the value
and the time.

Non-linear trend
When the time series data do not confirm with the linear trend, then we
obtain non-linear trend. We do so by obtaining a parabolic curve or non-
linear curve in the method of least squares. For this we use the equation of
the form.
  a  b  c 2  d 3 .......... k n which is known as a polynomial of
degree ‘n’ in ‘X’, k ≠ 0.
Let the parabolic curve be
  a  b   c 2
The values of a, b, and c can be determined by solving the normal
equations:

  ab  c  2
  a  b  2 c  3
 2   a  2 b  3  c  4
If we can change the origin at a suitable point, such that ‘X = 0’, then the
normal equations reduce to:

  ac  2
  b  2
 2   a  2 c  4
Manipal University Jaipur Page No. 510
Statistics for Management Unit 14

Self Assessment Questions

1. State ‘True’ or ‘False’
i) ‘The prices of cooking oils reduce after the harvesting of oil seeds
and go up after some time’ is an example of cyclic variations in a
time series.
ii) The effect of national strikes, floods, earthquakes are examples of
random variations in time series.

14.6 Mathematical Models for Time Series

The following are the two models commonly used for the decomposition of a
time series into its components.
 Additive model
 Multiplicative model
Most of the time series relating to economic and business phenomenon
conform to the multiplication model. In practice, additive model is rarely
used.
14.6.1 Additive Model

Key Statistic
The additive model assumes that the observed value is the sum of four
components of time series, that is,
Y=T+S+C+I
where,
Y = original data
T = trend value
S = seasonal component
C = cyclical component
I = irregular component
The additive model for decomposition of time series assumes that all the
four components of the time series operate independently of one another. It
also assumes that the behaviour of components is additive in character. It is
to be noted that only absolute values are added or deducted from the trend
value to arrive at the observed value.

Manipal University Jaipur Page No. 511

Statistics for Management Unit 14

14.6.2 Multiplicative model

Key Statistic
The multiplicative model assumes that the observed value is obtained by
multiplying the trend (T) by the rates of three other components, that is,
Y=TxSxCxI
where,
Y = original data
T = trend value
S = seasonal component
C = cyclical component
I = irregular component
The multiplicative model assumes that the components, although due to
different causes, are not necessarily independent and they can affect one
another. It also assumes that the behaviour of components is of
multiplicative character. It is to be noted that except the value of trend, all
the other values on the right hand side are rates or index numbers.

14.7 Editing of Time Series

It is necessary to make certain adjustments in the available data. Some
important adjustments are:
1. Time variation
When data is available on monthly basis, the effect of time variation needs
to be adjusted because all months of the year do not have the same number
of days. This adjustment of time variation is done by dividing each monthly
total by daily average. It is then multiplied by 365 / 12 which is the average
number of days in a month.
2. Population changes
Adjustment for population change becomes necessary when a variable is
affected by change in population. When we study national income figures,
such adjustment is necessary. In this case, adjustment is to divide the
income by the number of persons concerned. Then, we can have per capita
income figures.

Manipal University Jaipur Page No. 512

Statistics for Management Unit 14

3. Price changes
Adjustment for price changes becomes necessary wherever we have real
value changes. Current values are to be deflated by the ratio of current
prices to base year prices.
4. Comparability
The data which gets analysed should be comparable in order to have a valid
conclusion. When we deal with the analysis of time series it involves data
relating to the past which must be homogeneous and comparable.
Therefore, effects should be there to make the data as homogeneous and
comparable as possible.

14.8 Measurement of Seasonal Variation

In order to isolate and identify seasonal variations, we first eliminate the
effect of trend, cyclic variations and irregular fluctuations on the time series.
The main methods of measuring seasonal variations are:
 Simple average method
 Ratio to moving averages method
 Chain or link relative method
 Ratio to trend method
Now we will discuss separately each of the methods of measuring seasonal
variations.
14.8.1 Simple Average Method
In the simple average method, the steps followed are described below:
i) The time series is arranged by years and months or quarters.
ii) Totals of each month or quarter are obtained over all the years.
iii) The average for each month or quarter is obtained. The average may
be mean or median. In general, we take mean if not specified
otherwise.
iv) Taking the average of monthly or quarterly average equal to 100,
seasonal index for each month or quarter is calculated by the following
formula:
Seasonal Index for a month (or quarter) =

Monthly (or quarterly) average for the month (or quarter )

100
Average or monthly (or quarterly) averages

Manipal University Jaipur Page No. 513

Statistics for Management Unit 14

S
Symbolically, seasonal index for first term is given by: I  1
 100
1 S
Where, S1 = Average of first term
S = Average of all terms Sj / k where j = 1, 2, 3, 4……..k
k = 12 for monthly data
k = 4 for quarterly data
The merits and demerits of Simple average method are depicted in
table 14.8.
Table 14.9: Merits and Demerits of Simple Average Method
Merits Demerits
This method is the Most economic time series have trends and
simplest one. therefore, the seasonal index computed by this
method is really an index of trends and seasons.

This method is useful The simple averages method of isolating seasonal

where no definite trend fluctuations in time series is based on the
exists in the time series. assumption that the series contains only the
seasonal and irregular fluctuations.

This method does not give a true reflection of the

normal seasonal variation. This is because it is
obtained from the original data which is affected by
not only seasonal movements but also by
remaining three components.
The effects of cycles of the original data are not
eliminated by the process of averaging.

14.8.2 Ratio to moving averages method

Ratio to moving averages method is also known as percentage of moving
average method.
The steps involved in the computation of seasonal indices by this method
are described below:
i) The moving averages of the data are computed. If the data is monthly
then 12-monthly moving averages will be computed and if they are
quarterly then 4-quarterly moving averages will be computed. In both
the cases, time periods of moving averages are even. Hence, these
moving averages are to be centred.

Manipal University Jaipur Page No. 514

Statistics for Management Unit 14

ii) Under additive model, from each original value, the corresponding
moving average is deducted to find out short time fluctuations, which is
given as:
Y–T=S+C+I
iii) By preparing a separate table, monthly (or quarterly) short time
fluctuations are added for each month (or quarter) over all the years
and their average is obtained. These averages are known as seasonal
variations for each month or quarter.
iv) If we want to isolate / measure irregular variations, the mean of the
respective month or quarter is deducted from the short time
fluctuations.
14.8.3 Chain or link relative method
The steps involved in the chain or link relative method are described below.
i) Each quarterly or monthly value is divided by the preceding quarterly
or monthly value and the result is multiplied by 100. These
percentages are known as ‘Link Relatives’ of the seasonal values.
Thus:

Current Season Value

Link Re lative  100
Pr evious Season Value
There will be no link relative corresponding to the first.
ii) The mean of the link relatives for each season is computed over all the
years. Median can also be taken instead of mean of the link relatives.
iii) These average link relatives are converted into chain relatives. The
chain relative of first is taken as 100.
The Chain Re lative of current year


Average Link Re lative of current year  Chain Re lative of previous year 
100

iv) The second chain relative of first is computed on the basis of the chain
relative for the last:
The Chain Re lative of first quarter


Average Link Re lative of the first quarter  Chain Re lative of the last 
100

Manipal University Jaipur Page No. 515

Statistics for Management Unit 14

This chain relative may or may not be 100. It is not equal to 100 due
to secular trend. If it is 100, go to ‘step vi’, if it is not 100, go to ‘step
v’ and then go to ‘step vi’.
v) Compute the difference ‘d’ between the new chain relatives first
obtained in ‘step iv’ and chain relative assumed as 100. ‘d’ is divided
by the number of seasons and the resulting figure is multiplied by
1, 2, 3 and the product is deducted respectively from the chain
relatives of 2nd, 3rd, and 4th quarters. These are called corrected
relatives.
vi) The seasonal indices are obtained when the corrected chain relatives
are expressed as percentage of their relative averages.
14.8.4 Ratio to trend method
The following steps are considered to determine seasonal indices by this
method:
i) Determine the trend values by the method of least squares.
ii) To find ratio to trend, divide the original data by the corresponding
trend values and multiply these ratios by 100, that is,
 Original Data 
Ratio to Trend    100
 Trend Value 
iii) Calculate the arithmetic mean of the trend ratios obtained in ‘step ii’.
iv) Finally, all the trend ratios will be converted into seasonal indices. Add
all averages obtained in ‘step iii’ and find their general average.
Seasonal indices are calculated by using the following formula:
 Quarterly Averages 
Seasonal Indices    100
 General Averages 

14.9 Forecasting Methods Using Time Series

There are five forecasting methods using time series. They are:
1. Mean forecast
2. Naive forecast
3. Linear trend forecast
4. Non-linear trend forecast
5. Forecasting with exponential smoothing

Manipal University Jaipur Page No. 516

Statistics for Management Unit 14

14.9.1 Mean forecast

It is the simplest method of forecasting in which for the time period t, we
forecast the value of the series to be equal to the mean of the series, that is,

Y Y
t

In this method the trend effect and cyclic effects do not come into account.
14.9.2 Naive forecast
In this method we forecast the value, for the time period t, to be equal to the
actual value observed in the previous period, that is, time period (t-1). This
is given as:
Y Y
t t 1

14.9.3 Linear trend forecast

It is given by Yt = a + bX, where X is to be found from the value of t; a and b
are constants. This method is based on the least squares method where a
linear relationship is to be obtained between time and the response value ‘X’
by the formula which is given as:
Y Y
t t 1

14.9.4 Non-linear trend forecast

In this method a non-linear relationship between the time and the response
value has been found by the method of least squares. The value of forecast
‘Yt’ for the time period ‘t’, is given as:

Y  a  b  c 2 where, X-value will be calculated from the value of ‘t’

t

and the constant ‘a’.

14.9.5 Forecasting with exponential smoothing
Exponential smoothing is the forecasting method in which the observation
values are constantly updated and used to revise a forecast. As the
observations get older, they get exponentially decreasing weights.
Exponential smoothing is of many types, such as single, double, triple
exponential smoothing.

Manipal University Jaipur Page No. 517

Statistics for Management Unit 14

Self Assessment Questions

2. Fill in the following blanks.
i) A set of numerical value observed at regular interval of time is called
_______.
ii) Long term movements in time series are called ______.
iii) Variations that occur within a year are known as _______.
iv) Semi averages method is used to measure _________.
v) Method of moving averages does not show any _____ relationship.

Activity
Find seasonal variations by the ratio to trend method from the data given
below:

Year I Quarter II Quarter III Quarter IV Quarter

1994 60 80 72 68
1995 68 104 100 88
1996 80 116 108 96
1997 108 152 136 124
1998 160 184 172 164

Activity Solution
Quarterly
Yearly Trend
Year Average X X2 XY
Total Values
Y
1994 280 70 -2 4 140 64
1995 360 90 -1 1 -90 88
1996 400 100 0 0 0 112
1997 520 130 1 1 130 136
1998 680 170 2 4 340 160
∑Y = 560 ∑X= 0 ∑ X =10
2
∑XY= 240

Fitting of Linear Trend: Y = a + b X

The value of the constants a and b in the equation of a straight line is as
follows:
When,  X = 0 then normal equations reduce to:
Y 560
 Y  a , therefore, a
N
=
5
= 112

Manipal University Jaipur Page No. 518

Statistics for Management Unit 14

XY 240
 XY  b X 2
, therefore, b 
X 2

10
 24

Therefore, the equation will be given by: Y = 112 + 24X

Therefore, the quarterly increment is: (24/4)=6
Now quarterly trend values can be calculated:
Consider the year 1994, trend value is 64
Therefore the values for second and third Quarters of 1994 are: 64 - (6/2)
= (64 – 3)= 61 and 64 + (6/2) = (64+3) = 67 respectively. The value of the
first quarter of 1994 would be (61 – 6)= 55 and for the last quarter (67 +6)
= 73. Similarly all values are calculated.
Year I Quarter II Quarter III Quarter IV Quarter
1994 55 61 67 73
1995 79 85 91 97
1996 103 109 115 121
1997 127 133 139 145
1998 151 157 163 169
The given values of the time series will now be expressed as percentages
of the corresponding trend values given above. Thus, for the first quarter
of 1994, the percentage is (60/55)*100=109.1, second quarter
(80/61)*100= 131.1
Actual Quarterly Values as % of Quarterly Trend Values
Year I Quarter II Quarter III Quarter IV Quarter
1994 109.1 131.1 107.5 93.1
1995 86.1 122.4 109.9 90.7
1996 77.7 106.4 93.9 79.3
1997 85 114.3 97.8 85.5
1998 106.0 117.1 105.5 97.0
Total 463.9 591.3 514.6 445.6
Average 92.78 118.26 102.92 89.12
Seasonal
Index 92.0 117.4 102.1 88.4
Adjustment
Since the total of the seasonal index is 403.08=
(92.78+118.26+102.92+89.12) Each index has to be adjusted by
multiplying it by (400/403.08) and the final indices are thus available.

Manipal University Jaipur Page No. 519

Statistics for Management Unit 14

14.10 Summary
Let us recapitulate the important concepts discussed in this unit:
 A time series is a set of numerical values of a given variable listed at
successive intervals of time.
 The time series is classified into the following four components: Long
term trend or secular trend, Seasonal variations, Cyclic variations and
Random variations.
 The methods of measuring the trend of a time series are: Free hand or
graphic methods, Semi averages method, Moving average method and
Method of least squares.
 The forecasting methods using time series are: Mean forecast, Naïve
forecast, Linear trend forecast, Non-linear trend forecast and Forecasting
with exponential smoothing.

14.11 Glossary
Chain Relative: Ratio of seasonal index for the quarter to the seasonal
index of the previous quarter
Link Relative: Ratio of value of the variable for the quarter to the value of
the variable of the previous quarter.
Random: Changes in data value due to the factors other than trend and
seasonal.
Seasonal: Periodical changes in data values over regular intervals of time.
Time series: A set of observations recorded over a period of time.
Trend: The tendency in the data values either to increase or decrease.

14.12 Terminal Questions

1. What is meant by analysis of time series?
2. State the difference between seasonal variations and cyclical
fluctuations.
3. What is trend? State various methods of measuring it.
4. Explain the moving average method of measuring long term trend.
5. What are the components of time series? Bring out the significance of
moving average in analysing a time series and point out its limitations.

Manipal University Jaipur Page No. 520

Statistics for Management Unit 14

6. What is meant by secular trend? Discuss any two methods of isolating

trend values in a time series.
7. What is seasonal variation of a time series? Describe the various
methods you know to evaluate it and examine their relative merits.
8. Find a straight line trend to the following data and find trend value.
Table 14.10: Yearly Production Data
Year Production in 1000 (kg)
1990 80
1991 90
1992 92
1993 83
1994 94
1995 99
1996 92

9. Find seasonal values for the data in table 14.11.

Table 14.11: Data of Terminal Question 9
Year I Quarter II Quarter III Quarter IV Quarter
1995 3.7 4.1 3.3 3.5
1996 3.7 3.9 3.6 3.6
1997 4.0 4.1 3.3 3.1
1998 3.3 4.4 4.0 4.0

14.13 Answers

Self Assessment Questions

1. i) False
ii) True
2. i) Time series
ii) Secular trend
iii) Seasonal variations
iv) Trend
v) Functional

Manipal University Jaipur Page No. 521

Statistics for Management Unit 14

Terminal Questions
1. Refer section 14.2
2. Refer section 14.4.2 and section 14.4.3
3. Refer section 14.5
4. Refer section 14.5.3
5. Refer section 14.4 and section 14.5
6. Refer section 14.4
7. Refer section 14.8
8. The equation of the straight line is given as: Y = 90 + 2X
The trend values are 84, 86, 88, 90, 92, 94, 96.
9. The seasonal values obtained are 98.66, 110.74, 95.30, 95.30.

14.14 Case Study

The number of people visiting a hotel’s webpage per month is given below,
also called as cyber hits, for the year 2006-07 to 2008-09.The hotel could
forecast the future hits every month to facilitate its EDS department to plan
the server space, processing staff, etc.

Table 14.12: The number of people visiting a hotel’s webpage per month
Jan Feb Mar Apr May Jun July Aug Sep Oct Nov Dec
2006-07 420 100 300 344 300 200 344 766 899 900 788 455
2007-08 620 399 345 455 677 355 766 500 799 800 880 555
2008-09 520 289 400 644 566 677 500 800 899 900 680 666

Using the suitable methods with justification, forecast the number of visitors
to web page for all the month in the academic year Jan 2010-2011.

References:
 Agarwal B.L. (2006) Basic Statistics. 4th Ed., New Age International
Publishers.
 Anderson, David R., Sweeney. Dennis J. & Williams, Thomas A.
5th ed, Thomson Business Information Pvt Ltd.
 Bowerman, B. L. & O Connel, R.T. Applied Statistics: Improving
Business Processes. (1996) Irwin.
 Levin, Richard I. , Rubin, David S. (2008) Statistics for Management.
7th Ed. PHI Learning Private Limited.
Manipal University Jaipur Page No. 522
Statistics for Management Unit 14

 Pisani, Freedman D.R. & Purves, R. Statistics. (1997) 3rd Ed. W.W
Norton.
 Srivastava & Shailaja Rejo, T.N. (2008) Statistics for Management
5th Ed.TMH.
 Tanur, J.M. (2002) Statistics: A Guide to the unknown. 4th Ed. Brooks
/cole..
 Tukey, J.W. (1977) Exploratory Data Analysis. Addison –Wesley.
 Wilcox, Rand R. (2009) Basic Statistics – Understanding Conventional
Methods and Modern Insights. Oxford University Press.

E-References:
 http://www.textbooksonline.tn.nic.in/Books/11/Stat-EM/Chapter-1.pdf

Manipal University Jaipur Page No. 523