TIME SERIES ANALYSIS

Introduction to time series

Components of time series; Simple average,
Moving average

INTRODUCTION TO TIME SERIES ANALYSIS

This unit deals with time series methods. Every day, forecasting—the art
or science of predicting the future—is used in the decision making
process to help business people reach conclusions about buying, selling,
producing, hiring, and many other actions. As an example, consider the
following items:
■ Market watchers predict a resurgence of stock values next year.
■ City planners forecast a water crisis in Southern California.
■ Future brightens for solar power.
■ Energy secretary sees rising demand for oil.
■ CEO says difficult times won’t be ending soon for U.S. airline industry.
■ Life insurance outlook fades.
■ Increased competition from overseas businesses will result in
significant layoffs in the U.S. computer chip industry.

The emphasis in this unit is on time series analysis and forecasting. A

time series is a collection of data recorded over a period of time—weekly,
monthly, quarterly, or yearly.

How are these and other conclusions reached? What forecasting

techniques are used? Are the forecasts accurate? In this unit we discuss
several forecasting techniques. In addition, this unit will focus only on
data that occur over time, time-series data. Time-series data are data
gathered on a given characteristic over a period of time at regular
intervals. Time-series forecasting techniques attempt to account for
changes over time by examining patterns, cycles, or trends, or using
information about previous time periods to predict the outcome for a
future time period. Time-series methods include simple methods,
averaging, smoothing, regression trend analysis, and the decomposition of
the possible time-series factors.
Virtually all areas of business, including production, sales, employment,
transportation, distribution, and inventory, produce and maintain time-
series data.

The following section describes various components of time-series

analyses and to discuss alternative methods of economic and business
forecasting in terms of time-series data. First, a classical description of
four time-series components is offered. Then the moving average, Time
trend regression, exponential smoothing and forecasting.

COMPONENTS OF TIME SERIES

Several factors result in the interdependence of time-series data over time;

these factors are trend, seasonal, and business cycle factors. For example,
the current earnings of a growing company tend to be greater than its
earnings in the period just ended, and, of course, the expected earnings in
the next period will be greater than the current earnings. Therefore, the
correlation between any adjacent earnings is positive, and this is due to
the trend factor. Seasonal factors also contribute to the interdependence of
time-series data. Retail sales in the fourth quarter account for a major
portion of total annual sales of department stores. This seasonal factor
ensures that the sales volume in the fourth quarter of each year is highly
correlated with the fourth-quarter sales volume of any other year. The
business cycle is another cause of interdependency in a time-series model.

In short, it is traditionally assumed that the total variation in a time series

is composed of four basic components: a trend component, cyclical
component, a seasonal component and an irregular component. We will
now discuss these four components in some detail. Not all time-series
data have all these components.
 Figure – 1.1

TREND COMPONENT:
A trend is a pattern that exhibits a tendency either to grow or to
decrease fairly steadily over time. The long-term general direction of data
is referred to as trend. Notice that the data move through upward and
downward periods, the general direction or trend is increasing.
Figure - 1.2

CYCLE COMPONENT:
Cycles are patterns of highs and lows through which data move
over time periods usually of more than a year. Cyclical patterns are long-
term oscillatory patterns that are unrelated to seasonal behavior. They are
not necessarily regular but instead follow rather smooth patterns of
upswings and downswings, each swing lasting more than 2 or 3years
Time-series data that do not extend over a long period of time may not
have enough “history” to show cyclical effects.

Figure -1.3

SEASONAL COMPONENT:
The phenomenon of seasonality is common in the
business world. Seasonal effects, on the other hand, are shorter cycles,
which usually occur in time periods of less than one year. Often seasonal
effects are measured by the month, but they may occur by quarter, or may
be measured in as small a time frame as a week or even a day. Note the
seasonal effects have up and down cycles, many of which occur during a
1-year period. This seasonal behavior is quite clear and an obvious pattern
almost repeats itself each year.

Figure - 1.4
IRREGULAR COMPONENT:
The last component of the variation in a time series is the
irregular element introduced by the unexpected event. Irregular
fluctuations are rapid changes or “bleeps” in the data, which occur in even
shorter time frames than seasonal effects. Irregular fluctuations can
happen as often as day to day. They are subject to momentary change and
are often unexplained. For example, the announcement of a takeover bid
may cause the price of the target company’s stock to jump up 20 % or
more in a single day.

Figure -1.5

Table -1.1
Component Period length
Trend Long period
Insufficient data to observe
Cycle cyclical period
Season 4 quarters—yearly
No regular repeating pattern.
Irregular No period
SIMPLE AVERAGE

Many naïve model forecasts are based on the value of one time period.
Often such forecasts become a function of irregular fluctuations of the
data; as a result, the forecasts are “over estimated.” Using averaging
models, a forecaster enters information from several time periods into the
forecast and “smoothes” the data. Averaging models are computed by
averaging data from several time periods and using the average as the
forecast for the next time period.

The most elementary of the averaging models is the simple average

model. With this model, the forecast for time period t is the average of the
values for a given number of previous time periods, as shown in the
following.
Ft = (Xt-1 + Xt-2 + Xt-3 + ………..+ Xt-n)/n (1)

Example: Find the average number of students admitted in a particular

course for a 9 year data as follows:

Table - 1.2
Year(x) 2009 2010 2011 2012 2013 2014 2015 2016 2017
No: of students
admitted (y) 50 45 60 65 50 55 60 65 75

Solution: Calculation of the simple average of the number of students

admitted in a particular course

Table -1.3
No: of students
Year(x) admitted (y)
2009 50
2010 45
2011 60
2012 65
2013 50
2014 55
2015 60
 2016 65
2017 75
525

The simple average = 525/9 = 58.33

Advantage of simple averages

1. Fast and easy to calculate
2. Easy to work with and use in further analysis

Disadvantage of simple averages

1. Sensitive to extreme values
2. Not suitable for time series type of data
3. Works only when all values are equally important

MOVING AVERAGE

The most commonly used smoothing method is the method of moving

averages. A moving average replaces each value in a time series by an
average of the adjacent values. The number of values we use to construct
the averages is called the length of the moving average (L). Using
averaging models, a forecaster enters information from several time
periods into the forecast and “smoothes” the data. Averaging models are
computed by averaging data from several time periods and using the
average as the forecast for the next time period.

Example: Gross revenue data (Rupees in million) for a travel agency for
a 10 year period is as follows.

Table -1.4
Year (X) 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Revenue
(rupees in 3 6 10 8 7 12 14 14 18 19
million) (Y)

Calculate a 5-year moving average for the revenue earned.

Solution: Calculation for a 5 year moving average
Table -1.5
Revenue
(rupees in 5 year moving 5 year moving
Year (X) million) (Y) total average
2001 3
2002 6
2003 10 34 6.8
2004 8 43 8.6
2005 7 51 10.2
2006 12 55 11
2007 14 65 13
2008 14 77 15.4
2009 18
2010 19
For odd year (n+1)/2 moving average values are not available as shown in
Table 1.5

Figure -1.6 shows the actual line and a trend line

Example: Gross revenue data (Rupees in million) for a travel agency for
a 10 year period is as follows.
Table -1.6
Year (X) 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Revenue (rupees 3 6 10 8 7 12 14 14 18 19
in million) (Y)
Calculate a 3-year moving average for the revenue earned. Also find the
weighted 3-year moving average

Solution: Calculation for a 3 year moving average:

Table -1.7
Year Revenue 3 year 3 year Wt.Moving avg.
(x) (rupees in moving moving (w1=0.1,
million) (y) total average w2=0.4,w3=0.5)
2001 3
=(0.1*3+0.4*6+0.
2002 6 19 6.33 5*10) = 7.7
2003 10 24 8.00
2004 8 25 8.33
2005 7 27 9.00
2006 12 33 11.00
2007 14 40 13.33
2008 14 46 15.33
2009 18 51 17.00
2010 19

Figure - 1.7 shows the actual line and a trend line

20
18
16
14
12
10 Series1

8 Series2

6
4
2
0
2000 2002 2004 2006 2008 2010 2012

Advantages of moving averages

1. Moving averages can be used for measuring the trend of any series.
2. This method is applicable to linear as well as non-linear trends.
Disadvantages of moving averages

1. The trend obtained by moving averages generally is neither a

straight line nor a standard curve.
2. For this reason the trend cannot be extended for forecasting future
values.
3. Trend values are not available for some periods at the start and
some values at the end of the time series.
4. This method is not applicable to short time series.

LINEAR TREND METHOD

Introduction to Linear trend

linear trend method
Summary

INTRODUCTION TO LINEAR TREND

This section deals with linear trend. Developing the equation of a linear
trend line in forecasting is actually a special case of simple regression
where the y (or dependent variable) is the variable of interest that a
business analyst wants to forecast and for which a set of measurements
has been taken over a period of time. The long-term trend of many
business series, such as sales, exports, and production, often approximates
a straight line. The main use of the equation of a trend line by business
analysts is for forecasting outcomes for time periods in the future. Using a
regression model to predict y values for x values outside the domain of
those used to develop the model may not be valid. Despite this caution
and understanding the potential drawbacks, business forecasters
nevertheless extrapolate trend lines beyond the most current time periods
of the data and attempt to predict outcomes for time periods in the future.
To forecast for future time periods using a trend line, insert the time
period of interest into the equation of the trend line and solve for y .
LINEAR TREND EQUATION

yˆ  a  bx
(2)

where y hat, is the projected value of the Y variable for a selected value of
time x, a is the Y-intercept. It is the estimated value of Y when x = 0.
Another way to put it is: a is the estimated value of Y where the line
crosses the Y-axis when t is zero. b is the slope of the line, or the average
change in for each increase of one unit in x. x is any value of time that is
selected.

LINEAR TREND METHOD

After the equation of the line has been developed, several statistics are
available that can be used to determine how well the line fits the data.
Using the historical data values of x, predicted values of y (denoted as )
can be calculated by inserting values of x into the regression equation.
The predicted values can then be compared to the actual values of y to
determine how well the regression equation fits the known data. The
difference between a specific y value and its associated predicted y value
is called the residual or error of prediction.

In the field of forecasting, it is common to attempt to fit a trend line

through time-series data by determining the equation of the trend line and
then using the equation of the trend line to predict future data points.
To determine the equation of the regression line for a sample of data, the
researcher must determine the values for “a” and “b”. This process is
sometimes referred to as least squares analysis. Least squares analysis is a
process whereby a regression model is developed by producing the
minimum sum of the squared error values

yˆ  a  bx..............................................(3)
 y  na  b x..................................(4)
 xy  a x  b x^ 2.......................(5)
Example: Below are given the earnings (Rupees in lakhs). Fit a straight
line trend by the method of least squares to the given data. Assuming that
the same rate of change continues, what would be the predicted earnings
(Rupees in lakhs) for the year 2022?

Table -2.1
Earnings
(Rs. In
Year(x) lakhs) (Y)
2013 38
2014 40
2015 65
2016 72
2017 69
2018 60
2019 87
2020 95

Solution: The following table shows the calculations to fit a trend line to
the given data:
Table - 2.2
Earnings y=a+bx
(Rs. In Predicted
2
X Year(x) lakhs) (Y) XY X values
1 2013 38 38 1 40.1
2 2014 40 80 4 47.43
3 2015 65 195 9 54.76
4 2016 72 288 16 62.09
5 2017 69 345 25 69.42
6 2018 60 360 36 76.75
7 2019 87 609 49 84.08
8 2020 95 760 64 91.41
36 526 2675 204
By referring the equations (4) and (5)

From the table we have

526=8a+36b
2675=36a+204b

526=8a+36b *9
2675=36a+204b *2

4734=72a+324b
5350=72a+408b
616=84b
b=7.33

Substituting the value of b in equation (1) then we

get
a=32.77

Then we have y=32.77+7.33x

given x=10
Then we have y=32.77+7.33*10
y=106.065

The predicted values of Y are calculated using the values of ‘a’ and ‘b’
and are given in the Table -2.2
Figure -2.1 shows the actual line and the trend line
Advantages of Trend line

1. It is simple and easy to understand

2. It is used to forecasting and prediction

Disadvantages of Trend line

1. It is not always accurate

2. It assumes that forecast values of Y will be shaped by past pattern
only
***