Anda di halaman 1dari 31

Analysis of Time Series

Introduction
When the data is arranged
according to the time of their
occurrence, they form a time series.
Thus time series is the
arrangement of statistical
data in chronological order
daily, weekly, fortnightly,
monthly or yearly.

Components Of Time Series


When the data are arranged on the basis
of their occurrence often they show
fluctuations from time to time from day
to day, from week to week, from year to
year.
These fluctuations are caused
by a constantly working force.
This composite force has four
components, commonly
known as component of time
series, These forces are as
follows:

Secular trend or long term movements (T)


Periodic movements or short term
fluctuations
These comprise of
Seasonal Fluctuations (S)
Cyclic fluctuations (C)
Irregular or random
fluctuations (R)

Secular trend or long term movements (T)


Trend is that irreversible movement in time
series which continues in general in the
same direction over a long period of time.

Population,
Production,
Sales of
certain
products,
income,
prices

Deaths of
infants,
morality,
illiteracy

Trend refers to only smooth, regular, long term


movement of the data and has nothing to do with
sudden and erratic movements either in upward or
downward directions.

Uses of Trend

Business
Forecasting,
Planning of
future
Operations

Short
term and
irregular
movements
can be
studied

Two or
more series
can be
compared

Seasonal Fluctuations (S)


The periodic movements which occur regularly
every year and have their origin in the nature of the
year itself. Seasonal variations recur in regular and
periodic manner over a span of less than a year,
i.e., during a period of twelve months and have the
same almost the same pattern year after year.
Thus in a time series the data where only annual
figures are given, there are no seasonal variations.
Most of the time series relating to business and
economics are influenced by seasonal forces. e.g.,
time series relating to agriculture production, bank
clearings and deposits

These variations have two main causes:


Climate
In its
Widest
sense

Customs

Helps in
Future operations and formulating policy decisions
regarding purchase, production, Inventory control etc.

Cyclic fluctuations (C)


The wavelike moments in time series with period of
oscillation more than one year are known as cyclic
variations.
Cyclic variations generally exhibit semi regular
periodicity as they are neither as regular as seasonal
variations nor as accidental as erratic fluctuations.

Irregular or random
fluctuations (R)
It accounts for the random variability in the time
series. The irregular component is caused by short
term, unanticipated and non recurring forces that
affect the time series.
Random variations are unexpected variations both in
terms of amplitude and the length of time interval of
occurrence. These are forced by random forces and
sporadic causes.
Floods

Wars

lockouts

Storms

Models Of Decomposition
Analysis of time series involves the decomposition of
time series into its four components

There are two models of


Decomposition
Additive Model

Multiplicative model

Additive Model

The model is used when components are


independent of each other.
The magnitude of time series at any time is
given by:

Yt Tt C t St R t
The units of measurement are same for all
compnents.

When the time series data are recorded against years,


the seasonal components vanish and in the case:
Yt Tt C t R t

Multiplicative Model

This model is used it is assumed that forces


giving rise to the four type of variations are
independent
According to this model
Yt Tt C t St R t
log Yt log Tt logC t logSt logR t

Measures of Trend

Free hand
Or Graphical

Moving average

Semi average

Methods of
least square

Free hand Drawing


Years
1989 - 90
1990 - 91
1991 - 92
1992 - 93
1993 - 94
1994 - 95
1995 - 96
1996 - 97
1997 - 98
1998 - 99
1999 - 00

Imports
2005
2759
2041
1132
1355
3624
3399
2310
984
1608
3114

2000 - 01

2758

4000
3500
3000
2500
2000
1500
1000
500
0
1989 - 90 1990 - 91 1991 - 92 1992 - 93 1993 - 94 1994 - 95 1995 - 96 1996 - 97 1997 - 98 1998 - 99 1999 - 00 2000 - 01
Imports

Semi Average method


Steps to follow
Divide the series into two equal parts
Take average of each part separately
Plot the average of each part against the middle of
the time period covered by the respective parts
Join the plotted points

If the number of time periods is even, we have two


parts of the entire series easily but if the number of
time periods is odd, the normal practice is to ignore
the middle period and divide the resulting series in
to two halves.
Illustration:
Year

Sales (lkhs)

1991

38

1992

40

1993
1994

46
49

1995

51

1996

55

1997

61

1998

63

1999
2000

69
72

2001

80

Semi - total

Semi Average

224

44.8

345

69

90

80

70

60

50
sales

Sales (lkhs)

40

Linear (Sales (lkhs))

30

20

10

0
1990

1992

1994

1996
year

1998

2000

2002

Method of Moving Average


This method is based on the principle that the total
effect of periodic variations at different points of
time in its cycle gets completely neutralized in the
period of cyclic variations.
In this method we take average of a definite
number of terms (time period) in a time series in
succession, each time dropping one value at the
beginning and taking up the new value in sequence
from
the remaining
entries
in the series.
The period
of moving
average
say 3 years or 4
years or 5 years must coincide with the actual
period or multiples of the actual period of the
cyclical variations so that the cyclic variations are
eliminated.

This method helps in eliminating cyclic, seasonal


and random fluctuations in the time series data.
The periods of moving average can be even or odd.
When odd period of moving averages is used for
computing trend values moving totals and moving
averages are placed against the period of covered.
In this case the moving totals and moving averages
do not correspond to any year but fall between two
time periods, therefore the process of centering the
values is adopted. In this process two periods
moving average of the previously computed moving
averages is taken.

Example:
What is the trend in time series? The following table
gives the annual sales (in Rs 000) of a commodity
Year

Sales

Year

Sales

1990

710

1996

644

1991

705

1997

783

1992

680

1998

781

1993

687

1999

805

1994

757

2000

872

1995

629

Determine the trend by calculating 5 year moving


average.

Year

Sales

5 yearly moving average

5 yearly moving
average

1990

710

3539

707.8

1991

705

3458

691.6

1992

680

1993

687

3397

679.4

1994

757

3500

700

1995

629

3539

707.8

1996

644

1997

783

3594

718.8

1998

781

1999

805

3642

728.4

2000

872

3885

777

Method of Least - Squares


1. Fitting a straight line:
The equation is in the form Y = a + bX, where a;
trend value in the year of origin and b; Annual
change in the annual total (i.e. slope). By the
method of least squares. The normal equations to
find the values of a and b are.

Y na b X
XY a X b X

X is taken as time period and is treated as an


independent variable.
If the values of X are large, we redefine X by taking
deviations from some constant value i.e. we take x
= X A.

(i) If the number of time periods are odd, the


deviations are taken from the middle value, i.e.,
middle year which is (n+1)/2th value. So that x
becomes equal to 0 and the value of a is
determined from the equation.
Y

Y na b x a
n

And the value of b is determined from the equation.


xy

xY a x b x b x

(i) If the number of time periods are even then we


define,

X (middle

year )

1
( Interval between two middle years)
2

(i) If we have the years 1990, 1991, 1992, 1993,


1994, the middle year is 1992 and if the years are
1989, 1990, 1991, 1992, 1993, 1994, then two
middle years are 1991 and 1992. Middle value or
mean of two middle values = 1991.5 and
[ Interval between middle values or years] = 0.5.

Example:
Year

1992

1993

1994

1995

1996

Sales

12

18

20

23

27

Year
(X)

Sales (Y)

X 1994
(x)

xY

x2

Trend
Yc

1992

12

-2

-24

13.0

1993

18

-1

-18

16.5

1994

20

20.0

1995

23

23

23.5

1996

27

54

17.0

Total

N=5

100

35

10

Solution

Equation of straight line = Y = a + b (X 1994)


Since

x 0,
Y
a
b

N
xY
x2

100
20
5

35
3.5
10

and
Yc a b ( X 1994) 20 3.5 x
X 1992
Yc 20 3.5 2 13

Likely sales in 2001 is 44.5.

Anda mungkin juga menyukai