Spreadsheet Modeling

& Decision Analysis

A Practical Introduction to
Management Science
5th edition

Cliff T. Ragsdale

Chapter 11
Time Series Forecasting

Introduction to Time Series Analysis

A time-series is a set of observations on a quantitative
variable collected over time.
Examples
Dow Jones Industrial Averages
Historical data on sales, inventory, customer
counts, interest rates, costs, etc
Businesses are often very interested in forecasting
time series variables.
Often, independent variables are not available to build
a regression model of a time series variable.
In time series analysis, we analyze the past behavior
of a variable in order to predict its future behavior.

Some Time Series Terms

Stationary Data - a time series variable exhibiting
no significant upward or downward trend over
time.
Nonstationary Data - a time series variable
exhibiting a significant upward or downward trend
over time.
Seasonal Data - a time series variable exhibiting
a repeating patterns at regular intervals over
time.

Approaching Time Series Analysis

There are many, many different time series
techniques.
It is usually impossible to know which technique
will be best for a particular data set.
It is customary to try out several different
techniques and select the one that seems to
work best.
To be an effective time series modeler, you need
to keep several time series techniques in your
tool box.

Measuring Accuracy
We need a way to compare different time series techniques for a given data set.
Four common techniques are the:
mean absolute deviation,

mean absolute percent error,

Yi Y
i
n

Yi Yi

the mean square error,

root mean square error.

MAD =

i 1

100
MAPE =

n i 1

Yi

Yi Yi

MSE =
n
i 1
n

RMSE

We will focus on the

MSE.

MSE

Extrapolation Models
Extrapolation models try to account for the past
behavior of a time series variable in an effort to
predict the future behavior of the variable.

f Y , Y , Y ,
Y
t 1
t
t 1
t 2
Well first talk about several extrapolation
techniques that are appropriate for
stationary data.

An Example
Electra-City is a retail store that sells audio and
video equipment for the home and car.
Each month the manager of the store must order
merchandise from a distant warehouse.
Currently, the manager is trying to estimate how
many VCRs the store is likely to sell in the next
month.
He has collected 24 months of data.
See file Fig11-1.xls

Moving Averages
Yt Yt-1 Yt-k +1

Yt 1
k
No general method exists for determining k.
We must try out several k values to see what
works best.

Implementing the Model

See file Fig11-2.xls

A Comment on Comparing MSE Values

Care should be taken when comparing MSE
values of two different forecasting techniques.
The lowest MSE may result from a technique that
fits older values very well but fits recent values
poorly.
It is sometimes wise to compute the MSE using
only the most recent values.

Forecasting With
The Moving Average Model
Forecasts for time periods 25 and 26 at time
period 24:

Y24 Y23 36 + 35

Y25

355
.
2
2
Y
Y
35.5 + 36
25
24

Y26

35.75
2
2

Weighted Moving Average

The moving average technique assigns equal
weight to all previous observations
1
1
1

Yt 1 Yt Yt-1 Yt- k -1
k
k
k
The weighted moving average technique
allows for different weights to be assigned
to previous observations.
w Y w Y w Y
Y
t 1
1 t
2 t-1
k t- k -1
where 0 wi 1 and

We must determine values for k and the wi

Implementing the Model

See file Fig11-4.xls

Forecasting With The Weighted

Moving Average Model
Forecasts for time periods 25 and 26 at time period 24:

Y25 w1Y24 w2 Y23 0.291 36 0.709 35 35.29

Y26 w1Y25 w2 Y24 0.291 35.29 0.709 36 35.79

Exponential Smoothing
Y
(Y Y
)
Y
t 1
t
t
t
where 0 1
It can be shown that the above equation is
equivalent to:

Yt 1 Yt (1 )Yt 1 (1 ) 2 Yt 2 (1 ) n Yt n

Examples of Two
Exponential Smoothing Functions
42
40

Units Sold

38
36
34
32
Number of VCRs Sold

30

Exp. Smoothing alpha=0.1

Exp. Smoothing alpha=0.9

28
1

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Time Period

Implementing the Model

See file Fig11-8.xls

Forecasting With The

Exponential Smoothing Model
Forecasts for time periods 25 and 26 at time period 24:
Y
(Y Y
) 35.74 0.268(36 35.74) 35.81
Y
25
24
24
24
Y
(Y Y
)Y
(Y
Y
)Y
35.81
Y
26
25
25
25
25
25
25
25

Note that,
35.81, for t = 25, 26, 27,
Y
t

Seasonality
Seasonality is a regular, repeating pattern
in time series data.
May be additive or multiplicative in
nature...

Stationary Seasonal Effects

Additive Seasonal Effects

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

18

19

20

21

22

23

24

25

Tim e Period

Multiplicative Seasonal Effects

10

11

12

13

14

Tim e Period

15

16

17

Stationary Data With

Additive Seasonal Effects
where

Y
t n E t St n p
E t (Yt - St-p ) (1 - )E t 1
St (Yt - E t ) (1 - )St p
0 1

0 1
p represents the number of seasonal
periods

Et is the expected level at time period t.

St is the seasonal factor for time period t.

Implementing the Model

See file Fig11-13.xls

Forecasting With The Additive

Seasonal Effects Model
Forecasts for time periods 25 to 28 at time period 24:

Y
24 n E 24 S 24 n 4
E S 354.44 8.45 363.00
Y
25
24
21
E S 354.44 17.82 336.73
Y
26
24
22
E S 354.44 46.58 401.13
Y
27
24
23
E S 354.44 31.73 322.81
Y
28
24
24

Stationary Data With

Multiplicative Seasonal Effects
where

Y
t n E t St n p
E t (Yt /St-p ) (1 - )E t 1
St (Yt /E t ) (1 - )St p
0 1

0 1
p represents the number of seasonal periods

Et is the expected level at time period t.

St is the seasonal factor for time period t.

Implementing the Model

See file Fig11-16.xls

Forecasting With The Multiplicative

Seasonal Effects Model
Forecasts for time periods 25 to 28 at time period 24:

Y
24 n E 24 S 24 n 4
E S 353.95 1.015 359.13
Y
25
24
21
E S 354.44 0.946 334.94
Y
26
24
22
E S 354.44 1.133 400.99
Y
27
24
23
E S 354.44 0.912 322.95
Y
28
24
24

Trend Models
Trend is the long-term sweep or general
direction of movement in a time series.
Well now consider some nonstationary time
series techniques that are appropriate for
data exhibiting upward or downward trends.

An Example
WaterCraft Inc. is a manufacturer of personal
water crafts (also known as jet skis).
The company has enjoyed a fairly steady
growth in sales of its products.
The officers of the company are preparing sales
and manufacturing plans for the coming year.
Forecasts are needed of the level of sales that
the company expects to achieve each quarter.
See file Fig11-19.xls

Double Moving Average

Y
t n E t nTt
where

E t 2M t D t
Tt 2(M t D t ) /( k 1)
M t (Yt Yt 1 Yt k 1) / k
D t (Mt M t 1 Mt k 1) / k

Et is the expected base level at time period t.

Tt is the expected trend at time period t.

Implementing the Model

See file Fig11-20.xls

Forecasting With The

Double Moving Average Model
Forecasts for time periods 21 to 24 at time period 20:

Y
20 n E 20 nT20
E 1T 2385.33 1 139.9 2525.23
Y
21
20
20
E 2T 2385.33 2 139.9 2665.13
Y
22
20
20
E 3T 2385.33 3 139.9 2805.03
Y
23
20
20
E 4T 2385.33 4 139.9 2944.94
Y
24
20
20

Double Exponential Smoothing

(Holts Method)

Y
t n E t nTt
where
Et = Yt + (1-)(Et-1+ Tt-1)
Tt = (Et Et-1) + (1-) Tt-1
0 1 and 0 1

Et is the expected base level at time period t.

Tt is the expected trend at time period t.

Implementing the Model

See file Fig11-22.xls

Forecasting With Holts Model

Forecasts for time periods 21 to 24 at time period 20:

Y
20 n E 20 nT20

E 1T 2336.8 1 152.1 2488.9

Y
21
20
20
E 2T 2336.8 2 152.1 2641.0
Y
22
20
20
E 3T 2336.8 3 152.1 27931
Y
.
23

20

20

E 4T 2336.8 4 152.1 2945.2

Y
24
20
20

Holt-Winters Method For

Additive Seasonal Effects
Yt n E t nTt St n p
where

E t Yt St p (1 - )(E t 1 Tt 1 )

Tt E t E t 1 (1 - )Tt 1
St Yt E t (1 - )St p
0 1

0 1
0 1

Implementing the Model

See file Fig11-25.xls

Forecasting With Holt-Winters

Additive Seasonal Effects Method
Forecasts for time periods 21 to 24 at time period 20:

Y20 n E 20 nT20 S20 n 4

E 1 T S 2253.3 1 154.3 262.66 2670.3
Y
21
20
20
17
E 2 T S 2253.3 2 154.3 312.59 2249.3
Y
22
20
20
18
E 3 T S 2253.3 3 154.3 205.40 2921.6
Y
23
20
20
19
E 4 T S 2253.3 4 154.3 386.12 3256.6
Y
24
20
20
20

Holt-Winters Method For

Multiplicative Seasonal Effects
Yt n E t nTt St n p
where

E t Yt / St p (1 - )(E t 1 Tt 1 )

Tt E t E t 1 (1 - )Tt 1
St Yt / E t (1 - )St p
0 1

0 1
0 1

Implementing the Model

See file Fig11-28.xls

Forecasting With Holt-Winters

Multiplicative Seasonal Effects Method
Forecasts for time periods 21 to 24 at time period
20:

Y20 n E 20 nT20 S20 n 4

(E 1T ) S ( 2217.6 1 137.3)1.152 2713.7

Y
21
20
20 17
(E 2T ) S (2217.6 2 137.3)0.849 2114 .9
Y
22
20
20 18
(E 3T ) S (2217.6 3 137.3)1.103 2900.5
Y
23
20
20 19
(E 4T ) S (2217.6 4 137.3)1.190 3293.9
Y
24
20
20
20

The Linear Trend Model

Y t b0 b1X1t
where X1t t

For example:
X11 1, X12 2, X13 3,

Implementing the Model

See file Fig11-31.xls

Forecasting With
The Linear Trend Model
Forecasts for time periods 21 to 24 at time period 20:
b b X 3751
Y
. 92.6255 21 2320.3
21
0
1
121

b b X 3751
Y
. 92.6255 22 2412.9
22
0
1
122
b b X 3751
Y
. 92.6255 23 2505.6
23
0
1
123

b b X 3751
Y
. 92.6255 24 2598.2
24
0
1
124

The TREND() Function

TREND(Y-range, X-range, X-value for prediction)
where:
Y-range is the spreadsheet range containing the dependent
Y variable,
X-range is the spreadsheet range containing the
independent X variable(s),
X-value for prediction is a cell (or cells) containing the
values for the independent X variable(s) for which we want
an estimated value of Y.
Note: The TREND( ) function is dynamically updated whenever
any inputs to the function change. However, it does not provide
the statistical information provided by the regression tool. It is
best two use these two different approaches to doing regression
in conjunction with one another.

The Quadratic Trend Model

b b X b X
Y
t
0
1 1t
2 2t
where X1t t and X 2 t t 2

Implementing the Model

See file Fig11-34.xls

Forecasting With
The Quadratic Trend Model
Forecasts for time periods 21 to 24 at time period
20:

2
b b X b X
Y
21
0
1 121
2 2 21 653.67 16.671 21 3.617 21 2598.9
2
b b X b X
Y
22
0
1 122
2 2 22 653.67 16.671 22 3.617 22 2771.1
2
b b X b X
Y
23
0
1 123
2 2 23 653.67 16.671 23 3.617 23 2950.4
2
b b X b X
Y

653
.
67

16
.
671

24

3
.
617

24
3137.1
24
0
1 124
2 2 24

Computing Multiplicative
Seasonal Indices
We can compute multiplicative seasonal
adjustment indices for period p as follows:

Yi
i Y
i
Sp
, for all i occuring in season p
np
The final forecast for period i is then

adjusted = Y
S , for any i occuring in season p
Y
i
i
p

Implementing the Model

See file Fig11-37.xls

Forecasting With Seasonal Factors

Applied To The Quadratic Trend Model
Forecasts for time periods 21 to 24 at time period 20:
(b b X b X ) S 2598.9 105.7% 2747.8
Y
21
0
1 121
2 2 21
1
(b b X b X ) S 2771.1 80.1% 2219.6
Y
22
0
1 122
2 2 22
2
(b b X b X ) S 2950.5 103.1% 3041.4
Y
23
0
1 123
2 2 23
3

(b b X b X ) S 3137.2 111 .1% 3486.1

Y
24
0
1 124
2 2 24
4

Summary of the Calculation and Use

of Seasonal Indices
1. Create a trend model and calculate the estimated value (

for each Y
observation
in the sample.

2. For each observation, calculate the ratio of the actual value

to the predicted trend value:
.
Y /Y
(For additive effects, compute the difference:

).
Yt Y
t

3. For each season, compute the average of the ratios

calculated in step 2. These are the seasonal indices.

4. Multiply any forecast produced by the trend model by the

appropriate seasonal index calculated in step 3. (For additive
seasonal effects, add the appropriate factor to the forecast.)

Refining the Seasonal Indices

Note that Solver can be used to simultaneously
determine the optimal values of the seasonal
indices and the parameters of the trend model
being used.
There is no guarantee that this will produce a
better forecast, but it should produce a model
that fits the data better in terms of the MSE.
See file Fig11-39.xls

Seasonal Regression Models

Indicator variables may also be used in regression
models to represent seasonal effects.
If there are p seasons, we need p -1 indicator variables.

Our example problem involves quarterly data,

so p=4 and we define the following 3 indicator
variables: 1, if Yt is an observation from quarter 1
X 3t

0, otherwise

1, if Yt is an observation from quarter 2

X4t
0, otherwise
1, if Yt is an observation from quarter 3
X5t
0, otherwise

Implementing the Model

The regression function is:
b b X b X b X b X b X
Y
t
0
1 1t
2 2t
3 3t
4 4t
5 5t
where X1t t and X 2 t t 2

See file Fig11-42.xls

Forecasting With The

Seasonal Regression Model
Forecasts for time periods 21 to 24 at time period 20:
Y21 824.471 17.319(21) 3.485(21) 2 86.805(1) 424.736(0) 123.453(0) 2638.5
Y22 824.471 17.319(22) 3.485(22) 2 86.805(0) 424.736(1) 123.453(0) 2467.7
Y23 824.471 17.319(23) 3.485(23) 2 86.805(0) 424.736(0) 123.453(1) 2943.2
Y24 824.471 17.319(24) 3.485(24) 2 86.805(0) 424.736(0) 123.453(0) 3247.8

Crystal Ball (CB) Predictor

CB Predictor is an add-in that simplifies
the process of performing time series
analysis in Excel.
A trial version of CB Predictor is available
on the CD-ROM accompanying this book.
For more information on CB Predictor see:
http://www.decisioneering.com
See file Fig11-46.xls

Combining Forecasts
It is also possible to combine forecasts to create a composite forecast.
Suppose we used three different forecasting methods on a given data
set.

Denote the predicted value of time period t

using each method as follows:
F1t , F2t , and F3t
We could create a composite forecast as
follows:

Yt b0 b1F1t b2 F2 t b3 F3t

End of Chapter 11