TimeSeries Models;
Measuring Forecast Error
MGS3100 Chapter 13
Forecasting
Forecasting Models
Forecasting
Techniques
Qualitative
Models
Time Series
Methods
Causal
Methods
Delphi
Method
Jury of Executive
Opinion
Sales Force
Composite
Consumer Market
Survey
Naive
Moving
Average
Weighted
Moving Average
Exponential
Smoothing
Trend Analysis
Seasonality
Analysis
Simple
Regression
Analysis
Multiple
Regression
Analysis
Multiplicative
Decomposition
Time Series Models
General Form: Y = T * C * S , where
T = Trend  long term movement of mean
C = (Business) Cycle  an upturn or downturn not
caused by seasonal variation; effect of the
economy
S = Seasonal Variation  repetitive pattern
observed over a specific time period
= Error (random variation)
Practical Forecast Form: = T * S
C is important, but difficult to forecast
Dont forecast an error!
A stationary time series
Linear trend time series
Linear trend and seasonality time series
Time
Time
series
value
Future
Components of a Time Series
Time Series: Stationary Models
Stationary Model Assumptions
Assumes item forecasted will stay steady over time (constant
mean; random variation only)
Techniques will smooth out shortterm irregularities
Forecast for period t+1 is equal to forecast for period t+k; the
forecast is revised only when new data becomes available.
Stationary Model Types
Nave Forecast
Moving Average
Weighted Moving Average
Exponential Smoothing
Stationary Time Series Models:
The Nave Model
Whatever happened
last period will
happen again this
time
The model is simple
and flexible
Provides a baseline
to measure other
models
Attempts to capture
seasonal factors at
the expense of
ignoring trend
data Monthly :
data Quarterly :
12
4
=
=
t t
t t
Y F
Y F
1
=
t t
Y F
or
Measures of Forecast Error
Bias  The arithmetic sum of the
errors
MAD  Mean Absolute Deviation
MAPE Mean Absolute Percentage
Error
Mean Square Error (MSE)  Similar
to simple sample variance
Standard Error  Standard deviation of
the sampling distribution (the square
root of the MSE)
Bias, MAD, and MAPE  typically
used for time series
) (
t t
F Y Error Forecast =
T F Y MAD
t t
T
t
/   /T  error forecast 
1
T
1 t
= =
= =
T Y F Y MAPE
t t t
T
t
/ ] /  [ 100
1
=
=
T F Y
Bias
t t
T
t
/ ) (
/T error) (forecast
1
T
1 t
=
=
=
=
T F Y
MSE
t t
T
t
/ ) (
/T  error forecast 
2
1
T
1 t
2
=
=
=
=
Nave Forecast
Wallace Garden Supply
Forecasting
Period
Actual
Value
Nave
Forecast Error
Absolute
Error
Percent
Error
Squared
Error
January 10 N/A
February 12 10 2 2 16.67% 4.0
March 16 12 4 4 25.00% 16.0
April 13 16 3 3 23.08% 9.0
May 17 13 4 4 23.53% 16.0
June 19 17 2 2 10.53% 4.0
July 15 19 4 4 26.67% 16.0
August 20 15 5 5 25.00% 25.0
September 22 20 2 2 9.09% 4.0
October 19 22 3 3 15.79% 9.0
November 21 19 2 2 9.52% 4.0
December 19 21 2 2 10.53% 4.0
0.818 3 17.76% 10.091
BIAS MAD MAPE MSE
Standard Error (Square Root of MSE) = 3.176619
Storage Shed Sales
Nave Forecast Graph
Wallace Garden  Naive Forecast
0
5
10
15
20
25
February March April May June July August September October November December
Period
S
h
e
d
s
Actual Value
Nave Forecast
The Moving Average Method
The forecast is the average of the last n
observations of the time series.
n
Y Y Y
F
n t t t
t
1 1
1
...
+
+
+ + +
=
Stationary Time Series Models:
Moving Averages
Moving Averages
Wallace Garden Supply
Forecasting
Period
Actual
Value ThreeMonth Moving Averages
January 10
February 12
March 16
April 13 10 + 12 + 16 / 3 = 12.67
May 17 12 + 16 + 13 / 3 = 13.67
June 19 16 + 13 + 17 / 3 = 15.33
July 15 13 + 17 + 19 / 3 = 16.33
August 20 17 + 19 + 15 / 3 = 17.00
September 22 19 + 15 + 20 / 3 = 18.00
October 19 15 + 20 + 22 / 3 = 19.00
November 21 20 + 22 + 19 / 3 = 20.33
December 19 22 + 19 + 21 / 3 = 20.67
Storage Shed Sales
Moving Averages Forecast
Wallace Garden Supply
Forecasting 3 period moving average
Input Data Forecast Error Analysis
Period Actual Value Forecast Error
Absolute
error
Squared
error
Absolute
% error
Month 1 10
Month 2 12
Month 3 16
Month 4 13 12.667 0.333 0.333 0.111 2.56%
Month 5 17 13.667 3.333 3.333 11.111 19.61%
Month 6 19 15.333 3.667 3.667 13.444 19.30%
Month 7 15 16.333 1.333 1.333 1.778 8.89%
Month 8 20 17.000 3.000 3.000 9.000 15.00%
Month 9 22 18.000 4.000 4.000 16.000 18.18%
Month 10 19 19.000 0.000 0.000 0.000 0.00%
Month 11 21 20.333 0.667 0.667 0.444 3.17%
Month 12 19 20.667 1.667 1.667 2.778 8.77%
Average 1.333 2.000 6.074 10.61%
Next period 19.667 BIAS MAD MSE MAPE
Actual Value  Forecast
Moving Averages Graph
Three Period Moving Average
0
5
10
15
20
25
1 2 3 4 5 6 7 8 9 10 11 12
Time
V
a
l
u
e
Actual Value
Forecast
Stability vs. Responsiveness
Should I use a 2period moving average or a
3period moving average?
The larger the n the more stable the forecast.
A 2period model will be more responsive to
change.
We dont want to chase outliers.
But we dont want to take forever to correct for
a real change.
We must balance stability with responsiveness.
The Weighted Moving Average Method
Historical values of the time series are assigned
different weights when performing the forecast
Stationary Time Series Models:
Weighted Moving Averages
1 t
F
+
= w
1
Y
t
+ w
2
Y
t1
+w
3
Y
t2
+ + w
n
Y
tn+1
Ew
i
= 1
Weighted Moving Average
Wallace Garden Supply
Forecasting
Period
Actual
Value Weights ThreeMonth Weighted Moving Averages
January 10 0.222
February 12 0.593
March 16 0.185
April 13 2.2 + 7.1 + 3 / 1 = 12.298
May 17 2.7 + 9.5 + 2.4 / 1 = 14.556
June 19 3.5 + 7.7 + 3.2 / 1 = 14.407
July 15 2.9 + 10 + 3.5 / 1 = 16.484
August 20 3.8 + 11 + 2.8 / 1 = 17.814
September 22 4.2 + 8.9 + 3.7 / 1 = 16.815
October 19 3.3 + 12 + 4.1 / 1 = 19.262
November 21 4.4 + 13 + 3.5 / 1 = 21.000
December 19 4.9 + 11 + 3.9 / 1 = 20.036
Next period 20.185
Sum of weights = 1.000
Storage Shed Sales
Weighted Moving Average
Wallace Garden Supply
Forecasting 3 period weighted moving average
Input Data Forecast Error Analysis
Period Actual value Weights Forecast Error
Absolute
error
Squared
error
Absolute
% error
Month 1 10 0.222
Month 2 12 0.593
Month 3 16 0.185
Month 4 13 12.298 0.702 0.702 0.492 5.40%
Month 5 17 14.556 2.444 2.444 5.971 14.37%
Month 6 19 14.407 4.593 4.593 21.093 24.17%
Month 7 15 16.484 1.484 1.484 2.202 9.89%
Month 8 20 17.814 2.186 2.186 4.776 10.93%
Month 9 22 16.815 5.185 5.185 26.889 23.57%
Month 10 19 19.262 0.262 0.262 0.069 1.38%
Month 11 21 21.000 0.000 0.000 0.000 0.00%
Month 12 19 20.036 1.036 1.036 1.074 5.45%
Average 1.988 6.952 6.952 10.57%
Next period 20.185 BIAS MAD MSE MAPE
Sum of weights = 1.000
Stationary Time Series Models:
Exponential Smoothing
Exponential Smoothing
Moving average technique that requires a minimum
amount of past data
Uses a smoothing constant with a value between 0 and 1
(Usual range 0.1 to 0.3)
Forecast for period t = Forecast for period t1 plus times
the difference between the actual value and forecast in
period t1:
t
=
t1
+ (Y
t1

t1
), or
Can also be expressed as:
t
= (Y
t1
) + (1 )(
t1
) =
(Actual value in period t1) + (1 )(Forecast in period t1)
Exponential Smoothing Data
Period
Actual
Value(Y
t
)
t1
Y
t1
t1
t
January 10 = 10 0.1
February 12 10 + 0.1 *( 10  10 ) = 10.000
March 16 10 + 0.1 *( 12  10 ) = 10.200
April 13 10.2 + 0.1 *( 16  10.2 ) = 10.780
May 17 10.78 + 0.1 *( 13  10.78 ) = 11.002
June 19 11.002 + 0.1 *( 17  11.002 ) = 11.602
July 15 11.602 + 0.1 *( 19  11.602 ) = 12.342
August 20 12.342 + 0.1 *( 15  12.342 ) = 12.607
September 22 12.607 + 0.1 *( 20  12.607 ) = 13.347
October 19 13.347 + 0.1 *( 22  13.347 ) = 14.212
November 21 14.212 + 0.1 *( 19  14.212 ) = 14.691
December 19 14.691 + 0.1 *( 21  14.691 ) = 15.322
Storage Shed Sales
Class Exercise: What is the forecast for January of the following year?
How about March? Find the Bias, Mad & MAPE. (Note: equals 0.1.)
Exponential Smoothing
(Alpha = .419)
Wallace Garden Supply
Forecasting Exponential smoothing
Input Data Forecast Error Analysis
Period Actual value Forecast Error
Absolute
error
Squared
error
Absolute
% error
Month 1 10 10.000
Month 2 12 10.000 2.000 2.000 4.000 16.67%
Month 3 16 10.838 5.162 5.162 26.649 32.26%
Month 4 13 13.000 0.000 0.000 0.000 0.00%
Month 5 17 13.000 4.000 4.000 16.000 23.53%
Month 6 19 14.675 4.325 4.325 18.702 22.76%
Month 7 15 16.487 1.487 1.487 2.211 9.91%
Month 8 20 15.864 4.136 4.136 17.106 20.68%
Month 9 22 17.596 4.404 4.404 19.391 20.02%
Month 10 19 19.441 0.441 0.441 0.194 2.32%
Month 11 21 19.256 1.744 1.744 3.041 8.30%
Month 12 19 19.987 0.987 0.987 0.973 5.19%
Average 2.608 9.842 14.70%
Alpha 0.419 MAD MSE MAPE
Next period 19.573
Exponential Smoothing
Exponential Smoothing
0
5
10
15
20
25
J
a
n
u
a
r
y
F
e
b
r
u
a
r
y
M
a
r
c
h
A
p
r
i
l
M
a
y
J
u
n
e
J
u
ly
A
u
g
u
s
t
S
e
p
t
e
m
b
e
r
O
c
t
o
b
e
r
N
o
v
e
m
b
e
r
D
e
c
e
m
b
e
r
S
h
e
d
s
Actual value
Forecast
Evaluating the Performance
of Forecasting Techniques
Several forecasting methods have been
presented.
Which one of these forecasting methods
gives the best forecast?
Time 1 2 3 4 5 6
Time series: 100 110 90 80 105 115
3Period Moving average: 100 93.33 91.6
Error for the 3Period MA:  20 11.67 23.4
3Period Weighted MA(.5, .3, .2) 98 89 85.5
Error for the 3Period WMA  18 16 29.5
Performance Measures
Sample Example
Find the forecasts and the errors for each forecasting
technique applied to the following stationary time series.
MAD for the moving average technique:
MAD for the weighted moving average technique:
= 21.17
= 18.35
20 + 11.67 + 23.4
3
MAD = =
E
,A
t

n
18 + 116 + 29.5
3
MAD = =
E
,A
t

n
Performance Measures
MAD for the Sample Example
MAPE for the moving average technique:
MAPE for the weighted moving average technique:
= .211
= .188
20/80 + 11.67/105+ 23.4/115
3
MAPE= =
E
,A
t

n
18/80 + 16/105 + 29.5/115
3
MAPE= =
E
,A
t

n
Performance Measures
MAPE for the Sample Example
Use the performance measures to select a good set
of values for each model parameter.
For the moving average:
the number of periods (n).
For the weighted moving average:
The number of periods (n),
The weights (w
i
).
For the exponential smoothing:
The exponential smoothing factor (o).
Excel Solver can be used to determine the values
of the model parameters.
Performance Measures
Selecting Model Parameters
Trend & Seasonality
Trend analysis
Technique that fits a trend equation (or curve) to a
series of historical data points
Projects the equation into the future for medium and
long term forecasts. Typically do not want to forecast
into the future more than half the number of time
periods used to generate the forecast
Seasonality analysis
Adjustment to time series data due to variations at
certain periods.
Adjust with seasonal index  ratio of average value of
the item in a season to the overall annual average value.
Examples: demand for coal in winter months; demand
for soft drinks in the summer and over major holidays
Linear Trend Analysis
Midwestern Manufacturing Sales
Scatter Diagram
Actual
value (or)
Y
Period
number
(or) X
74 1995
79 1996
80 1997
90 1998
105 1999
142 2000
122 2001
Sales(in units) vs. Time
0
20
40
60
80
100
120
140
160
1994 1995 1996 1997 1998 1999 2000 2001 2002
Least Squares for Linear Regression
Midwestern Manufacturing
Least Squares Method
Time
V
a
l
u
e
s
o
f
D
e
p
e
n
d
e
n
t
V
a
r
i
a
b
l
e
s
Objective: Minimize
the squared deviations!
Least Squares Method
bX a Y
^
+ =
Where
Y
^
= predicted value of the dependent variable (demand)
a = Yaxis intercept =  b*
b = Slope of the regression line =
] X n  XY [
_ _
Y
_
2 2
X n  X
X = value of the independent variable (time)
X Y
Linear Trend Data & Error Analysis
Midwestern Manufacturing Company
Forecasting Linear trend analysis
Input Data Forecast Error Analysis
Period
Actual value
(or) Y
Period number
(or) X Forecast Error
Absolute
error
Squared
error
Absolute
% error
Year 1 74 1 67.250 6.750 6.750 45.563 9.12%
Year 2 79 2 77.786 1.214 1.214 1.474 1.54%
Year 3 80 3 88.321 8.321 8.321 69.246 10.40%
Year 4 90 4 98.857 8.857 8.857 78.449 9.84%
Year 5 105 5 109.393 4.393 4.393 19.297 4.18%
Year 6 142 6 119.929 22.071 22.071 487.148 15.54%
Year 7 122 7 130.464 8.464 8.464 71.644 6.94%
Average 8.582 110.403 8.22%
Intercept 56.714 MAD MSE MAPE
Slope 10.536
Next period 141.000 8
Enter the actual values in cells shaded YELLOW. Enter new time period at the bottom to forecast
Least Squares Graph
Trend Analysis
y = 10.536x + 56.714
0
20
40
60
80
100
120
140
160
1 2 3 4 5 6 7
Time
V
a
l
u
e
Actual values Linear (Actual values)
Run linear regression to test 
1
in the model
Y
t
=
0
+
1
t+c
t
Excel results:
Coeff. Stand. Err t Stat P value Lower 95%Upper 95%
Intercept 369.27 27.79436 13.2857 5E18 313.44 425.094
Weeks 0.3339 0.912641 0.36586 0.71601 1.49919 2.16699
0.71601
This large Pvalue indicates
that there is little evidence that trend exists
Conclusion: A stationary model is appropriate.
Another way to Determine Trend:
Use the Excel Regression Function
Forecasting Seasonal Data: Quick Method
Eichler Supplies
Year Month Demand
Average
Demand Ratio
Seasonal
Index
1 January 80 94 0.851 0.957
February 75 94 0.798 0.851
March 80 94 0.851 0.904
April 90 94 0.957 1.064
May 115 94 1.223 1.309
June 110 94 1.170 1.223
July 100 94 1.064 1.117
August 90 94 0.957 1.064
September 85 94 0.904 0.957
October 75 94 0.798 0.851
November 75 94 0.798 0.851
December 80 94 0.851 0.851
2 January 100 94 1.064 0.957
February 85 94 0.904 0.851
March 90 94 0.957 0.904
April 110 94 1.170 1.064
May 131 94 1.394 1.309
June 120 94 1.277 1.223
July 110 94 1.170 1.117
August 110 94 1.170 1.064
September 95 94 1.011 0.957
October 85 94 0.904 0.851
November 85 94 0.904 0.851
December 80 94 0.851 0.851
Seasonal Index ratio of the
average value of the item in a
season to the overall average
annual value.
Example: average of year 1
January ratio to year 2 January
ratio.
(0.851 + 1.064)/2 = 0.957
Ratio = Demand / Average Demand
If Year 3 average monthly demand is
expected to be 100 units.
Forecast demand Year 3 January:
100 X 0.957 = 96 units
Forecast demand Year 3 May:
100 X 1.309 = 131 units
Forecasting Seasonal Data With Trend
1. Calculate the seasonal indices (as shown on the
previous slide)
2. Calculate deseasonalized treand by dividing
the actual value (Y) by the seasonal index for
that period:
Deseasonalized Trend = Y / Seasonal index
(e.g., 80 units/ 0.957 = 83.595)
3. Find the trend line, and extend the trend line into
the desired forecast period.
Forecasting Seasonal Data With Trend:
Calculating the Seasonal Forecast
4. Now that we have the Seasonal Indices and Trend
line, we can reseasonalize the data and generate
the seasonalized forecast by multiplying the
trend line values in the forecast period by the
appropriate seasonal indices for each time period
as follows:
= Trend x Seasonal Index