Forecasting
Subject Expert:
Roy B. Gacus
1
Intended Learning
Outcome
• By the end of the learning
experience, students must be
able to:
• Use the traditional forecasting
methods such as time series
analysis and regression to
provide reliable guidelines for
decision making.
2
Chapter Topics
Forecasting Components
Time Series Methods
Forecast Accuracy
Regression Methods
3
Forecasting Components
A variety of forecasting methods are available for use
depending on the time frame of the forecast and the
existence of patterns.
Time Frames:
Short-range (one to two months)
Medium-range (two months to one or two years)
Long-range (more than one or two years)
Patterns:
Trend
Random variations
Cycles
Seasonal pattern
4
Forecasting Components
Patterns (1 of 2)
Trend - A long-term movement of the item being forecast.
Random variations - movements that are not predictable
and follow no pattern.
Cycle - A movement, up or down, that repeats itself over a
lengthy time span.
Seasonal pattern - Oscillating movement in demand that
occurs periodically in the short run and is repetitive.
5
Forecasting Components
Patterns (2 of 2)
Figure 15.1
Forms of Forecast Movement: (a) Trend, (b) Cycle, (c) Seasonal Pattern,
(d) Trend with Seasonal Pattern
6
Forecasting Components
Forecasting Methods
Time Series - Statistical techniques that use historical data
to predict future behavior.
Regression Methods - Regression (or causal ) methods that
attempt to develop a mathematical relationship between the
item being forecast and factors that cause it to behave the
way it does.
Qualitative Methods - Methods using judgment, expertise
and opinion to make forecasts.
7
Forecasting Components
Qualitative Methods
Qualitative methods, the “jury of executive opinion,” is the
most common type of forecasting method for long-term
strategic planning.
Performed by individuals or groups within an organization,
sometimes assisted by consultants and other experts,
whose judgments and opinion are considered valid for the
forecasting issue.
Usually includes specialty functions such as marketing,
engineering, purchasing, etc. in which individuals have
experience and knowledge of the forecasted item.
Supporting techniques include the Delphi Method, market
research, surveys, etc.
8
Time Series Methods
Overview
Statistical techniques that make use of historical data
collected over a long period of time.
Methods assume that what has occurred in the past will
continue to occur in the future.
Forecasts based on only one factor - time.
9
Time Series Methods
Moving Average (1 of 5)
Moving average uses values from the recent past to
develop forecasts.
This dampens or smoothes out random increases and
decreases.
Useful for forecasting relatively stable items that do not
display any trend or seasonal pattern.
Formula for:
n
Di
MAn i1n
where:
n number of periods in the moving average
Di data in period i
10
Time Series Methods
Moving Average (2 of 5)
Example: Instant Paper Clip Supply Company forecast of
orders for the next month.
Three-month moving average:
3
Di
MA i1 90 110 130 110 orders
3 3 3
Five-month moving average:
5
Di
MA i1 90 110 130 75 50 91 orders
5 5 5
11
Time Series Methods
Moving Average (3 of 5)
Figure 15.2
Three- and Five-Month Moving Averages
12
Time Series Methods
Moving Average (4 of 5)
Figure 15.2
Three- and Five-Month Moving Averages
13
Time Series Methods
Moving Average (5 of 5)
Longer-period moving averages react more slowly to
changes in demand than do shorter-period moving
averages.
The appropriate number of periods to use often requires
trial-and-error experimentation.
Moving average does not react well to changes (trends,
seasonal effects, etc.) but is easy to use and inexpensive.
Good for short-term forecasting.
14
Time Series Methods
Weighted Moving Average (1 of 2)
In a weighted moving average, weights are assigned to the
most recent data.
Formula:
n
WMAn Wi Di
i1
where Wi the weight for period i, between 0% and 100%
Wi 1.00
Example: Paper clip company weights 50% for October, 33%
for September, 17% for August :
3
WMA Wi Di (.50)(90) (.33)(110) (.17)(130) 103.4 orders
3 i1
15
Time Series Methods
Weighted Moving Average (2 of 2)
Determining precise weights and number of
periods requires trial-and-error experimentation.
Chapter 15 - Forecasting 16
Time Series Methods
Exponential Smoothing (1 of 11)
Exponential smoothing weights recent past data more
strongly than more distant data.
Two forms: simple exponential smoothing and adjusted
exponential smoothing.
Simple exponential smoothing:
Ft + 1 = Dt + (1 - )Ft
where: Ft + 1 = the forecast for the next period
Dt = actual demand in the present period
Ft = the previously determined forecast for
the present period
= a weighting factor (smoothing constant).
17
Time Series Methods
Exponential Smoothing (2 of 11)
The most commonly used values of are between .10
and .50.
Determination of is usually judgmental and subjective and
often based on trial-and -error experimentation.
18
Time Series Methods
Exponential Smoothing (3 of 11)
Example: PM Computer Services (see Table 15.4).
Exponential smoothing forecasts using smoothing constant
of .30.
Forecast for period 2 (February):
F2 = D1 + (1- )F1 = (.30)(37) + (.70)(37) = 37 units
Forecast for period 3 (March):
F3 = D2 + (1- )F2 = (.30)(40) + (.70)(37) = 37.9
units
19
Time Series Methods
Exponential Smoothing (4 of 11)
Table 15.4
Exponential Smoothing Forecasts, = .30 and = .50
20
Time Series Methods
Exponential Smoothing (5 of 11)
The forecast that uses the higher smoothing constant (.50)
reacts more strongly to changes in demand than does the
forecast with the lower constant (.30).
Both forecasts lag behind actual demand.
Both forecasts tend to be consistently lower than actual
demand.
Low smoothing constants are appropriate for stable data without
trend; higher constants appropriate for data with trends.
21
Time Series Methods
Exponential Smoothing (6 of 11)
Figure 15.3
Exponential Smoothing Forecasts
22
Time Series Methods
Exponential Smoothing (7 of 11)
Adjusted exponential smoothing: exponential smoothing with a trend adjustment factor
added.
Formula:
AFt + 1 = Ft + 1 + Tt+1
where: T = an exponentially smoothed trend factor
Tt + 1 = (Ft + 1 - Ft) + (1 - )Tt
Tt = the last period trend factor
= smoothing constant for trend ( a value between zero
and one).
Reflects the weight given to the most recent trend data.
Determined subjectively.
Chapter 15 - Forecasting 23
Time Series Methods
Exponential Smoothing (8 of 11)
Example: PM Computer Services exponential smoothed
forecasts with = .50 and = .30 (see Table 15.5).
Adjusted forecast for period 3:
T3 = (F3 - F2) + (1 - )T2
= (.30)(38.5 - 37.0) + (.70)(0) = 0.45
AF3 = F3 + T3 = 38.5 + 0.45 = 38.95
24
Time Series Methods
Exponential Smoothing (9 of 11)
Table 15.5
Adjusted Exponentially Smoothed Forecast Values
25
Time Series Methods
Exponential Smoothing (10 of 11)
Adjusted forecast is consistently higher than the simple
exponentially smoothed forecast.
It is more reflective of the generally increasing trend of the
data.
26
Time Series Methods
Exponential Smoothing (11 of 11)
Figure 15.4
Adjusted Exponentially Smoothed Forecast
27
Time Series Methods
Linear Trend Line (1 of 5)
When demand displays an obvious trend over time, a least
squares regression line , or linear trend line, can be used to
forecast.
Formula:
y a bx b xy n x y
where : x2 n x 2
a intercept (at period 0) a y b x
b slope of the line where :
x the time period n number of periods
y forecast for demand x nx
for period x
y ny
28
Time Series Methods
Linear Trend Line (2 of 5)
Table 15.6
Least Squares Calculations
29
Time Series Methods
Linear Trend Line (3 of 5)
Example: PM Computer Services (see Table 15.6)
x 78 6.5 y 557 46.42
12 12
b xy n x y 3,867 (12)(6.5)(46.42) 1.72
x 2 n x 2 650 12(6 . 5)2
a y b x 46.42 (1.72)(6.5) 35.2
y 35.2 1.72 x linear trend line
for period 13, x 13, y 35.2 1.72(13) 57.56
30
Time Series Methods
Linear Trend Line (4 of 5)
A trend line does not adjust to a change in the trend as
does the exponential smoothing method.
This limits its use to shorter time frames in which trend will
not change.
31
Time Series Methods
Linear Trend Line (5 of 5)
Figure 15.5
Linear Trend Line
32
Regression Methods
Overview
Time series techniques relate a single variable being
forecast to time.
Regression is a forecasting technique that measures the
relationship of one variable to one or more other variables.
Simplest form of regression is linear regression.
33
Regression Methods
Linear Regression
Linear regression relates demand (dependent variable ) to
an independent variable.
y a bx
a y b x
b xy n x y
x 2 n x 2
where :
x nx mean of the x data
y ny mean of the y data
34
Regression Methods
Linear Regression Example (1 of 3)
State University athletic department.
Wins Attendance x y
4 36,300 (wins) (attendance, 1,000s) xy x2
6 40,100 4 36.3 145.2 16
6 41,200 6 40.1 240.6 36
8 53,000 6 41.2 247.2 36
6 44,000 8 53.0 424.0 64
7 45,600 6 44.0 264.0 36
5 39,000 7 45.6 319.2 49
7 47,500 5 39.0 195.0 25
7 47.5 332.5 49
49 346.7 2,167.7 311
35
Regression Methods
Linear Regression Example (2 of 3)
x 49 6.125
8
y 346 .9 43.34
8
b xy n x y (2,167 .70 (8)(6.125)(43.34) 4.06
x 2 n x 2 (311 ) (8)(6 . 125)2
a y b x 43.34 (.406)(6.125) 18.46
Therefore, y 18.46 4.06 x
Attendance forecast for x 7 wins is
y 18.46 4.06(7) 46.88 or 46,880
36
Regression Methods
Linear Regression Example (3 of 3)
Figure 15.6
Linear Regression Line
37
Multiple Regression with Excel (1 of 4)
Multiple regression relates demand to two or more
independent variables.
General form:
y = 0 + 1x1 + 2x2 + . . . + kxk
where 0 = the intercept
1 . . . k = parameters representing
contributions of the independent
variables
x1 . . . xk = independent variables
Chapter 15 - Forecasting 38
Multiple Regression with Excel (2 of 4)
State University example:
Wins Promotion ($) Attendance
4 29,500 36,300
6 55,700 40,100
6 71,300 41,200
8 87,000 53,000
6 75,000 44,000
7 72,000 45.600
5 55,300 39,000
7 81,600 47,500
Chapter 15 - Forecasting 39
Multiple Regression with Excel (3 of 4)
Exhibit 15.16
Chapter 15 - Forecasting 40
Multiple Regression with Excel (4 of 4)
Exhibit 15.17
Chapter 15 - Forecasting 41
Correlation Methods (1 of 2)
Correlation is a measure of the strength of the relationship
between independent and dependent variables.
Pearson r correlation coefficient formula:
r n xy x y
2
2
2 2
n
x x n y y
Value lies between +1 and -1.
Value of zero indicates little or no relationship between
variables.
Values near 1.00 and -1.00 indicate strong linear
relationship.
42
Correlation Methods (1 of 2)
Value for State University example:
r (8)(2,167.7) (49)(346.7) .948
(8)(311) (49)(49) (8)(15,224.7) (346.7)2
43
Coefficient of Determination
The Coefficient of determination is the percentage of the
variation in the dependent variable that results from the
independent variable.
Computed by squaring the correlation coefficient, r.
For State University example:
r = .948, r2 = .899
This value indicates that 89.9% of the amount of variation in
attendance can be attributed to the number of wins by the
team, with the remaining 10.1% due to other, unexplained,
factors.
44
Time Series Methods
Seasonal Adjustments (1 of 4)
A seasonal pattern is a repetitive up-and-down movement
in demand.
Seasonal patterns can occur on a monthly, weekly, or daily
basis.
A seasonally adjusted forecast can be developed by
multiplying the normal forecast by a seasonal factor.
A seasonal factor can be determined by dividing the actual
demand for each seasonal period by total annual demand:
Si =Di/D
45
Time Series Methods
Seasonal Adjustments (2 of 4)
Seasonal factors lie between zero and one and represent
the portion of total annual demand assigned to each
season.
Seasonal factors are multiplied by annual demand to
provide adjusted forecasts for each period.
46
Time Series Methods
Seasonal Adjustments (3 of 4)
Example: Wishbone Farms
Table 15.7
Demand for Turkeys at Wishbone Farms
S1 = D1/ D = 42.0/148.7 = 0.28
S2 = D2/ D = 29.5/148.7 = 0.20
S3 = D3/ D = 21.9/148.7 = 0.15
S4 = D4/ D = 55.3/148.7 = 0.37
47
Time Series Methods
Seasonal Adjustments (4 of 4)
Multiply forecasted demand for entire year by seasonal
factors to determine quarterly demand.
Forecast for entire year (trend line for data in Table):
y = 40.97 + 4.30x = 40.97 + 4.30(4) = 58.17
Seasonally adjusted forecasts:
SF1 = (S1)(F5) = (.28)(58.17) = 16.28
SF2 = (S2)(F5) = (.20)(58.17) = 11.63
SF3 = (S3)(F5) = (.15)(58.17) = 8.73
SF4 = (S4)(F5) = (.37)(58.17) = 21.53
48
Forecast Accuracy
Overview
Forecasts will always deviate from actual values.
Difference between forecasts and actual values referred to as forecast
error.
Would like forecast error to be as small as possible.
If error is large, either technique being used is the wrong one, or
parameters need adjusting.
Measures of forecast errors:
Mean Absolute deviation (MAD)
Mean absolute percentage deviation (MAPD)
Cumulative error (E bar)
Average error, or bias (E)
Chapter 15 - Forecasting 49
Forecast Accuracy
Mean Absolute Deviation (1 of 7)
MAD is the average absolute difference between the
forecast and actual demand.
Most popular and simplest-to-use measures of forecast
error.
Formula:
Dt Ft
MAD n
where:
t the period number
Dt demand in period t
Ft the forecast for period t
n the total number of periods
50
Forecast Accuracy
Mean Absolute Deviation (2 of 7)
Example: PM Computer Services (see Table 15.8).
Compare accuracies of different forecasts using MAD:
Dt Ft 53.41
MAD n 4.85
11
51
Forecast Accuracy
Mean Absolute Deviation (3 of 7)
Table 15.8
Computational Values for MAD
52
Forecast Accuracy
Mean Absolute Deviation (4 of 7)
The lower the value of MAD relative to the magnitude of the
data, the more accurate the forecast.
When viewed alone, MAD is difficult to assess.
Must be considered in light of magnitude of the data.
53
Forecast Accuracy
Mean Absolute Deviation (5 of 7)
Can be used to compare accuracy of different forecasting
techniques working on the same set of demand data (PM
Computer Services):
Exponential smoothing ( = .50): MAD = 4.04
Adjusted exponential smoothing ( = .50, = .30):
MAD = 3.81
Linear trend line: MAD = 2.29
Linear trend line has lowest MAD; increasing from .30
to .50 improved smoothed forecast.
54
Forecast Accuracy
Mean Absolute Deviation (6 of 7)
A variation on MAD is the mean absolute percent deviation
(MAPD).
Measures absolute error as a percentage of demand rather
than per period.
Eliminates problem of interpreting the measure of accuracy
relative to the magnitude of the demand and forecast
values.
Formula:
Dt Ft 53.41
MAPD .103 or 10.3%
Dt 520
55
Forecast Accuracy
Mean Absolute Deviation (7 of 7)
MAPD for other three forecasts:
Exponential smoothing ( = .50): MAPD = 8.5%
Adjusted exponential smoothing ( = .50, = .30):
MAPD = 8.1%
Linear trend: MAPD = 4.9%
56
Forecast Accuracy
Cumulative Error (1 of 2)
Cumulative error is the sum of the forecast errors (E =et).
A relatively large positive value indicates forecast is biased
low, a large negative value indicates forecast is biased high.
If majority of errors are positive, forecast is consistently low;
and vice versa.
Cumulative error for trend line is always almost zero, and is
therefore not a good measure for this method.
Cumulative error for PM Computer Services can be read
directly from Table 15.8.
E = et = 49.31 indicating forecasts are frequently below
actual demand.
Chapter 15 - Forecasting 57
Forecast Accuracy
Cumulative Error (2 of 2)
Cumulative error for other forecasts:
Exponential smoothing ( = .50): E = 33.21
Adjusted exponential smoothing ( = .50, =.30):
E = 21.14
Average error (bias) is the per period average of cumulative
error.
Average error for exponential smoothing forecast:
E net 49 .31 4.48
11error indicates a forecast is
A large positive value of average
biased low; a large negative error indicates it is biased high.
58
Forecast Accuracy
Example Forecasts by Different Measures
Table 15.9
Comparison of Forecasts for PM Computer Services
Results consistent for all forecasts:
Larger value of alpha is preferable.
Adjusted forecast is more accurate than exponential
smoothing forecasts.
Linear trend is more accurate than all the others.
59