19- 1

Time Series Analysis

Md. Siddikur Rahman, PhD

Associate Professor

Department of Statistics
Begum Rokeya University, Rangpur
 19- 2

Time Series and Forecasting

GOALS
When you have completed this lecture, you will be able to:
ONE
Define the four components of a time series.
TWO
Compute a moving average.
THREE
Determine a linear trend equation.
FOUR
Compute the trend equation for a nonlinear trend.
 19- 3

Time Series and Forecasting

GOALS
When you have completed this lecture, you will be able to:

FIVE
Use trend equations to forecast future time periods and
to develop seasonally adjusted forecasts.
SIX
Determine and interpret a set of seasonal indexes.
SEVEN
Deseasonalize data using a seasonal index.
 19- 4

A Time Series is a collection of data recorded over a

period of time. The data may be recorded weekly,
monthly, or quarterly.

There are four The Secular Trend

components to
a time series: The Cyclical Variation
The Seasonal Variation

The Irregular Variation

Components of a Time Series
 19- 5

The Cyclical Variation is the rise and fall of a time

series over periods longer than one year.
25-Year Sales of Real Estate

140
120
Sales in $(Millions)

100
80
60
40
20
0
1975 1980 1985 1990 1995 2000 2005

Year

Cyclical Variation
 19- 6

The Secular Trend is the smooth long run direction of

the time series.

Six-year sales by Season

16
14
12
Sales in $(millions)

10
8
6
4
2
0
r er ll er er ll er er ll er er ll er er ll er er ll
te ng Fa int pr in
g
Fa int prin
g
Fa int prin
g
Fa int prin
g
Fa int prin
g
Fa
in pr i m m m m m m m
W S Su W S Sum W S Sum W S Sum W S Sum W S Sum
Seasons 1999-2005

The Secular Trend

 19- 7

The Seasonal Variation is the pattern of change in a

time series within a year. These patterns tend to repeat
themselves from year to year.
Six-year sales by Season

16
14
Sales in $(millions)

12
10
8
6
4
2
0
g r l g r l g r l g r l g r l g r l
n ter rin me Fal nter rin me Fal nter rin me Fal nter rin me Fal nter rin me Fal nter rin me Fal
i p i p i p i p i p i p
W S Sum W S Sum W S Sum W S Sum W S Sum W S Sum
Seasons 1999-2005

Seasonal Variation
 19- 8

The Irregular Variation is divided into two

components:

Episodic Residual
Variations variations
are random in
are nature and
unpredictable, cannot be
but can usually identified.
be identified,
such as a flood or
hurricane. Components of a Time
Series
 19- 9

The long term trend equation (linear) :

Y´ = a + bt

o Y’ is the projected value of the Y variable for a

selected value of t

o a is the Y-intercept, the estimated value of Y

when t=0.

o b is the slope of the line

o t is an value of time that is selected Linear Trend

 19- 10

The owner of Strong Homes would like a forecast

for the next couple of years of new homes that
will be constructed in the Pittsburgh area. Listed
below are the sales of new homes constructed in
the area for the last 5 years.

Year Sales ($1000)

1999 4.3
2000 5.6
2001 7.8
2002 9.2
2003 9.7
Example 1
 19- 11

Regression Statistics
Multiple R 0.982
R Square 0.964
Adjusted R Square 0.952 EXCEL
Standard Error 0.507 Regression tool
Observations 5

ANOVA
df SS MS F
Regression 1 20.736 20.736 80.580
Residual 3 0.772 0.257
Total 4 21.508

Coefficients Standard Error t Stat P-value

Intercept 3.00 0.532 5.639 0.011
X Variable 1 1.44 0.160 8.977 0.003

Example 1 Continued
 19- 12

12
y = 1.44x + 3
10
EXCEL
Chart 8

and 6
trend
4
line
function 2

0
1 2 3 4 5

Example 1 continued
 19- 13

The same results

can be derived
using MINITAB’s The time series equation
Stat/time series/ is:
trend analysis Y’ = 3.00 + 1.44t
function.

The forecast for the year

2003 is:
Y´ = 3.00 + 1.44(7) = 13.08

Example 1 continued
 19- 14

If the trend is not linear but rather the

increases tend to be a constant percent, the Y
values are converted to logarithms, and a least
squares equation is determined using the logs.

log(Y ') = [log(a )] + [log(b)]t

Nonlinear Trends
 19- 15

Technological advances are so rapid that often

initial prices decrease at an exponential rate from
month to month. Hi-Tech Company provides the
following information for the 12-month period
after releasing its latest product.
Example 2
 19- 16
Month Unit Price Month Unit Price
1 900 13 208
2 786 14 187
3 687 15 168
4 603 16 152
5 531 17 137
6 468 18 125
7 414 19 113
8 367 20 103
9 326 21 94
10 290 22 86
11 259 23 78
12 232 24 72

Example 2 continued
 19- 17

Price of new product: First 24 months

1000

800

600 y = -31.454x + 700.99

Price

400

200

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-200
Time period

Example 2 continued
 19- 18

Month Unit Price Log Price Month Unit Price Log Price
1 900 2.954 13 208 2.318
2 786 2.895 14 187 2.272
3 687 2.837 15 168 2.226
4 603 2.781 16 152 2.182
5 531 2.725 17 137 2.138
6 468 2.670 18 125 2.096
7 414 2.617 19 113 2.054
8 367 2.565 20 103 2.013
9 326 2.513 21 94 1.972
10 290 2.463 22 86 1.933
11 259 2.414 23 78 1.894
12 232 2.366 24 72 1.856

Example 2 continued
 19- 19

Price of New Product: First 24 Months

Using logs of price values

3.5
3
Log (Unit Price)

y = -0.0476x + 2.9596
2.5
2
1.5
1
0.5
0
1 3 5 7 9 11 13 15 17 19 21 23

Time period

Example 2 continued
 19- 20

Take antilog to find

estimate. Thus, the
estimated sales price
for the 25th period
would be:

Price25 = antilog(-.0476*25+2.9596)

= 58.830

Example 2 continued
 19- 21

The Moving-Average method is used to smooth out a

time series. This is accomplished by “moving” the
arithmetic mean through the time series.

To apply the The moving-average is the basic

moving-average method used in measuring the
method to a time seasonal fluctuation.
series, the data
should follow a
fairly linear trend
and have a
definite rhythmic
pattern of The Moving-Average
Method
fluctuations.
 19- 22

The method most commonly used to compute the typical

seasonal pattern is called the Ratio-to-Moving-
Average method.

It eliminates the The numbers that

trend, cyclical, result are called the
and irregular Typical
components from
the original data
Seasonal
(Y). Indexes.

Seasonal Variation
 19- 23

Using an example of sales in a large toy company, let us look

at the steps in using the moving average method.

Winter Spring Summer Fall

1998 6.7 4.6 10.0 12.7
1999 6.5 4.6 9.8 13.6
2000 6.9 5.0 10.4 14.1
2001 7.0 5.5 10.8 15.0
2002 7.1 5.7 11.1 14.5
2003 8.0 6.2 11.4 14.9
Determining a Seasonal Index
 19- 24

Step 1: Determine the moving total for the time series.

Step 2: Determine the moving average for the time
series.
Step 3: The moving averages are then centered.
Step 4: The specific seasonal for each period is then
computed by dividing the Y values with the centered
moving averages.
Step 5: Organize the specific seasonals in a table.
Step 6: Apply the correction factor.

Steps
 19- 25

The resulting series (sales) is called deseasonalized

sales or seasonally adjusted sales.
The reason for deseasonalizing
a series (sales) is to remove the
seasonal fluctuations so that
the trend and cycle can be
studied.
A set of typical indexes is very
useful in adjusting a series
(sales, for example)

Deseasonalizing Data
 19- 26
Deseasonalized Toy Sales

Winter Spring Summer Fall

1998 8.7 8.0 8.7 8.3
1999 8.5 8.0 8.6 8.9
2000 9.0 8.7 9.1 9.2
2001 9.1 9.5 9.4 9.8
2002 9.3 9.9 9.7 9.5
2003 10.4 10.8 10.0 9.8

Example 3
 $Sales (Millions)
W

0.0
2.0
4.0
6.0
8.0
10.0
12.0
in
Sp ter
Su ring
m
m
er
F
W al
in
Sp ter
Su ring
m
m
er
F
W al
in
Sp ter
Su ring
m
m
er
W alF
in
Sp ter
Su ring
m
Quarter
m
er
Fa
W l
in
Deseasonalized Toy Sales

Sp ter
Su ring
m
m
Deseasonalized Toy Sales

er
F
W al
in
Sp ter
Su ring
m
m
er
Fa
l
Example 3 continued
19- 27