4 - 1© 2014 pearson education, inc. forecasting powerpoint presentation to accompany heizer and...
TRANSCRIPT
4 - 1© 2014 Pearson Education, Inc.
Forecasting
PowerPoint presentation to accompany Heizer and Render Operations Management, Eleventh EditionPrinciples of Operations Management, Ninth Edition
PowerPoint slides by Jeff Heyl
44
© 2014 Pearson Education, Inc.
4 - 2© 2014 Pearson Education, Inc.
Outline
▶ What Is Forecasting?
▶ Seven Steps in the Forecasting System
▶ Forecasting Approaches▶ Time-Series Forecasting
▶ Associative Forecasting Methods: Regression and Correlation Analysis
▶ Accuracy of Forecasts
4 - 3© 2014 Pearson Education, Inc.
WHAT IS FORECASTING?
4 - 4© 2014 Pearson Education, Inc.
What is Forecasting?► Process of predicting a
future event
► Underlying basis of all business decisions► Production
► Inventory
► Personnel
► Facilities
??
4 - 5© 2014 Pearson Education, Inc.
1. Short-range forecast► Up to 1 year, generally less than 3 months
► Purchasing, job scheduling, workforce levels, job assignments, production levels
2. Medium-range forecast► 3 months to 3 years
► Sales and production planning, budgeting
3. Long-range forecast► 3+ years
► New product planning, facility location, research and development
Forecasting Time Horizons
Comprehensive
Accurate
4 - 6© 2014 Pearson Education, Inc.
Distinguishing Differences
1. Medium/long range forecasts deal with more comprehensive issues and support management decisions regarding planning and products, plants and processes
2. Short-term forecasting usually employs different methodologies than longer-term forecasting
3. Short-term forecasts tend to be more accurate than longer-term forecasts
4 - 7© 2014 Pearson Education, Inc.
Demand PatternsQ
uan
tity
Time
(a) Stable pattern: Data cluster about a horizontal line
4 - 8© 2014 Pearson Education, Inc.
Demand PatternsQ
uan
tity
Time
(b) Trend: Data consistently increase or decrease
4 - 9© 2014 Pearson Education, Inc.
Demand PatternsQ
uan
tity
| | | | | | | | | | | |J F M A M J J A S O N D
Months
(c) Seasonal: Data consistently show regular up-and-down movements in a time series that relate to recurring events such as weather or holidays
Year 1
Year 2
4 - 10© 2014 Pearson Education, Inc.
Seven Steps in Forecasting1. Determine the use of the forecast
2. Select the items to be forecasted
3. Determine the time horizon of the forecast
4. Select the forecasting model(s)
5. Gather the data needed to make the forecast
6. Make the forecast
7. Validate and implement results
4 - 11© 2014 Pearson Education, Inc.
The Realities!
► Forecasts are seldom perfect, unpredictable outside factors may impact the forecast
► Most techniques assume an underlying stability in the system
► Product family and aggregated forecasts are more accurate than individual product forecasts
4 - 12© 2014 Pearson Education, Inc.
FORECASTING APPROACHES
4 - 13© 2014 Pearson Education, Inc.
Forecasting Approaches
ForecastingMethods
Qualitative
Jury of executive opinion
Market surveyDelphi method
Sales force composite
Quantitative
Naïve approach
Average Technique
Time-seriesmodel
Associative forecasting method
Trend projection
Exponential smoothing
Simple moving average
Weighted moving average
4 - 14© 2014 Pearson Education, Inc.
Qualitative Forecasting Approach
► Used when situation is vague and little data exist► New products
► New technology
► Involves intuition, experience► e.g., forecasting sales on Internet
Qualitative Methods
4 - 15© 2014 Pearson Education, Inc.
▶Executive Opinions▶Involve small group of high-level managers
▶Group estimates demand by working together
▶Combine managerial experience with statistical models
▶Relatively quick
▶‘Group-think’ disadvantage: The view of one person may prevail
Qualitative Forecasting Approach
4 - 16© 2014 Pearson Education, Inc.
▶Market Survey▶Ask customers about purchasing plans
▶What consumers say, and what they actually do are often different
▶Surveys can be expensive and time-consuming
▶Low response rate to a mail survey
▶May be overly optimistic
Qualitative Forecasting Approach
4 - 17© 2014 Pearson Education, Inc.
▶Delphi method▶Iterative group process, continues until
consensus is reached
▶Responses are kept anonymous
▶Long-term forecast
▶Applied to a variety of situation, including forecast
Qualitative Forecasting Approach
4 - 18© 2014 Pearson Education, Inc.
▶Sales force Composite▶Sales staff are good source of information
because of their direct contact with consumer.
▶Each salesperson projects his or her sales
▶Combined at district and national levels
▶May be overly influenced by recent experiences
▶Tends to be overly optimistic
Qualitative Forecasting Approach
4 - 19© 2014 Pearson Education, Inc.
Forecasting Approaches
ForecastingMethods
Qualitative
Jury of executive opinion
Market surveyDelphi method
Sales force composite
Quantitative
Naïve approach
Average Technique
Time-seriesmodel
Associative forecasting method
Trend projection
Exponential smoothing
Simple moving average
Weighted moving average
4 - 20© 2014 Pearson Education, Inc.
Forecasting Approaches
► Used when situation is ‘stable’ and historical data exist► Existing products► Current technology
► Involves mathematical techniques► e.g., forecasting sales of color
televisions
Quantitative Methods
4 - 21© 2014 Pearson Education, Inc.
Time-series Forecasting
Time Market Demand1/1/2008 601/2/2008 501/3/2008 801/4/2008 70
1/11/2008 4201/18/2008 4001/25/2008 470
Not evenly spaced data
► Set of evenly spaced numerical data► Forecast based only on past values, no
other variables important► Assumes that factors influencing past and
present will continue influence in future
4 - 22© 2014 Pearson Education, Inc.
Time-series Forecasting
[Solution]: Yes, they do represent a time series since an observation was made each quarter for a period of six quarters or 1.5 years.
Year Quarter Period number Sales (yards)
2012 1 1 16,000
2 2 18,500
3 3 12,500
4 4 15,000
2013 1 5 13,500
2 6 17,500
Example 4.1: The table below shows the carpet sales in 2012 and the first half of 2013 at Hoppy Toppy in Addison, TX. Do the quarterly sales represent a time series? Why or why not?
4 - 23© 2014 Pearson Education, Inc.
Time-series Forecasting
▶Notation: t = Index of time period
At = Actual demand in period t, t = 1, 2, ...
Ft = Forecast for period t, t = 1, 2, ...
wt = Weight assigned to period t, t = 1, 2, …
4 - 24© 2014 Pearson Education, Inc.
Naive Approach
► Assumes demand in next period is the same as demand in most recent period► e.g., If January sales were 68, then
February sales will be 68
► Sometimes cost effective and efficient► Can be good starting point
Ft = At-1
4 - 25© 2014 Pearson Education, Inc.
Naive Approach Example 4.2: The table below shows the computers sales from 2007 to 2013. Apply the naïve approach to predict how many computers will be needed in 2014.
F8= A7 = 35,000
Year Time Period Sales
2007 1 25,000
2008 2 27,000
2009 3 30,000
2010 4 32,000
2011 5 28,000
2012 6 29,000
2013 7 35,000
2014 8 ?
the most recent period
4 - 26© 2014 Pearson Education, Inc.
Simple Moving Average (SMA)
▶Used if little or no trend
▶ The SMA forecast is equal to the average value of the observations in the most recent n periods, where n is the number of time periods involved and its value is determined by the decision maker
n
AAAF nt2t1t
t
Or,
© 2014 Pearson Education, Inc. 4 - 27
F8 = (A7 + A6 + A5) / 3 = (35,000 + 29,000 + 28,000) / 3 = 30,700
the 3 most recent years
Simple Moving Average (SMA)
Year Time Period Sales
2007 1 25,000
2008 2 27,000
2009 3 30,000
2010 4 32,000
2011 5 28,000
2012 6 29,000
2013 7 35,000
2014 8 ?
Example 4.3: Assuming that n=3, apply three-year moving average approach to predict how many computers will be needed in 2014.
4 - 28© 2014 Pearson Education, Inc.
Weighted Moving Average (WMA)
▶ Weights based on experience and intuition
▶ Older data less important: heavier weights are assigned to more recent observations, wt-1 wt-2 … wt-n.
nt2t1t
ntnt2t2t1t1tt w...ww
Aw...AwAwF
Weighted moving average
Or,
© 2014 Pearson Education, Inc. 4 - 29
Weighted Moving Average (WMA)
the 4 most recent years
Year Time Period Sales
2007 1 25,000
2008 2 27,000
2009 3 30,000
2010 4 32,000
2011 5 28,000
2012 6 29,000
2013 7 35,000
2014 8 ?
Example 4.4: Assuming that n=4 and appropriate weights are 4, 3, 2 and 1, apply the WMA approach to predict how many computers will be needed in 2014
500,311234
1320002280003290004350008
F
4 - 30© 2014 Pearson Education, Inc.
Moving Average (SMA, WMA)▶ More time periods are involved, the smoother the forecasts
will be. So, the moving average approach is less sensitive to real changes in the data.
4 - 31© 2014 Pearson Education, Inc.
Moving Average (SMA, WMA)
▶Moving averages lag the actual value.
▶Moving averages require extensive records of past data.
4 - 32© 2014 Pearson Education, Inc.
Exponential Smoothing (ES)
New forecast = Last period’s forecast+ (Last period’s actual demand
– Last period’s forecast)
Ft = Ft – 1 + (At – 1 - Ft – 1)
where Ft = new forecast
Ft – 1 = previous period’s forecast
= smoothing (or weighting) constant (0 ≤ ≤ 1)
At – 1 = previous period’s actual demand
Example 4.5: In January, a car dealer predicted February demand for 142 Ford Mustangs. Actual February demand was 153 autos. Using a smoothing constant chosen by management of = 0.2, apply the ES approach to forecast March demand.
Previous forecast = 142
Previous actual = 153
= 0.2
Current forecast = Previous forecast + (Previous Actual – Previous forecast)= 142 + 0.2(153 - 142)= 144.2
Thus, the March demand forecast is rounded to 144.
Exponential Smoothing (ES)
4 - 34© 2014 Pearson Education, Inc.
EX 1 in classA dry cleaner uses exponential smoothing to forecast equipment usage at its main plant. August usage was forecasted to be 88 percent of capacity; actual usage was 89.6 percent of capacity. A smoothing constant of 0.1 is used. Prepare a forecast for September.
4 - 35© 2014 Pearson Education, Inc.
Exponential Smoothing (ES)▶ The naive forecast is a special case of the ES with = 1.
▶ Large values of are chosen when the underlying average is likely to change. Small values of are used when the underlying average is fairly stable.
▶ As increases, older values become less significant
Ex 2 in classSuppose the manager of an ice cream store is trying to forecast the pounds of ice cream that they will sell based on what they have sold in the past four days. Recent actual sales numbers are:
Day Sales (in lbs)Sunday 137.1Monday 123.6Tuesday 134.9Wednesday 160.0Thursday 140.4
1) What is a four-day moving average forecast of Friday’s demand?
2) Let’s assume the following weights. What is a four-day weighted moving average forecast of Friday’s demand?
3) Let’s say that actual sales for a given day totaled 115 pounds, while the forecast for that day was 110 pounds. With a smoothing constant of 0.1, what is the next day’s forecast?
Day Weight4 days ago 0.13 days ago 0.22 days ago 0.2Yesterday 0.5Total 1.0
4 - 37© 2014 Pearson Education, Inc.
Forecasting Approaches
ForecastingMethods
Qualitative
Jury of executive opinion
Consumer market surveyDelphi method
Sales force composite
Quantitative
Naïve approach
Average Technique
Time-seriesmodel
Associative forecasting method
Trend projection
Exponential smoothing
Simple moving average
Weighted moving average
4 - 38© 2014 Pearson Education, Inc.
Comparison
▶Averaging Techniques -- when the time series generally varies around an average.
▶Trend Projection Method -- when the time series generally varies around a linear trend (i.e., upward and downward).
4 - 39© 2014 Pearson Education, Inc.
Trend Projection
Upward Linear Trend Downward Linear Trend
4 - 40© 2014 Pearson Education, Inc.
Trend Projection
btay ˆ
Suppose that a linear trend is present in the time series. The main goal of trend projection analysis is searching for a linear equation that best describes the trend. The best-fit trend line may be represented by the following equation:
y = Computed value of the variable to be predicted (called the dependent variable)
t = Specified number of time periods (called the independent variable)
b = Slope of the trend line (or the rate of change in y for given changes in t)
22 tnt
ytnty
a = Intercept of the trend line at the vertical axis tby n = number of data points or observations
© 2014 Pearson Education, Inc. 4 - 41
Year Period Demand
2003 1 74
2004 2 80
2005 3 82
2006 4 74
2007 5 90
2008 6 100
2009 7 86
2010 8 94
2011 9 112
2012 10 104
2013 11 110
Example 4.6: Data on the demand for JM's computers have been accumulated for 11 years and are shown in the following table:
(1) Plot the time-series data to determine the most appropriate forecasting approach
65
75
85
95
105
115
Demand (y)
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013year (x)
An upward linear trend is present in the time series. The trend projection approach is the most appropriate forecasting method.
© 2014 Pearson Education, Inc. 4 - 42
t y t2 ty
1 74
2 80
3 82
4 74
5 90
6 100
7 86
8 94
9 112
10 104
11 110
66 1006
1
4
9
16
25
36
49
64
81
100
121
506
74
160
246
296
450
600
602
752
1008
1040
1210
6438
(2) How many computers will be needed in each of the coming two years?
Sum
The best-fit trend line is y = 69.49 + 3.66t
2014: F12 = 69.49 + 3.66(12) = 113.41 or about 113 units
2015: F13 = 69.49 + 3.66(13) = 117.07 or about 117 units
66.3611506
45.916116438222
tnt
ytntyb
611
66
n
tt 45.91
11
1006
n
yy
49.69666.345.91 tbya
4 - 43© 2014 Pearson Education, Inc.
EX3 in class
Room registrations in the Toronto Towers Plaza Hotel have been recorded for the past 9 years. To project future occupancy, management would like to determine the mathematical trend of guest registration. This estimate will help the hotel determine whether future expansion will be needed. Given the following time-series data, develop a trend equation to forecast 2014 registrations. Room registrations are in the thousands:
Year 2005 2006 2007 2008 2009 2010 2011 2012 2013Registrations(in thousands)
17 16 16 21 20 20 23 25 24
4 - 44© 2014 Pearson Education, Inc.
Forecasting Approaches
ForecastingMethods
Qualitative
Jury of executive opinion
Consumer market surveyDelphi method
Sales force composite
Quantitative
Naïve approach
Average Technique
Time-seriesmodel
Associative forecasting method
Trend projection
Exponential smoothing
Simple moving average
Weighted moving average
4 - 45© 2014 Pearson Education, Inc.
Associative Forecasting
▶ Used when changes in one or more independent variables can be used to predict the changes in the dependent variable
▶Dependent variable
▶We wish to predict
▶Example: demand for doorknobs
▶ Independent variable
▶Affect the dependent variable and thereby “cause” the results observed in the past.
▶Example: advertising expenditures and new housing starts
4 - 46© 2014 Pearson Education, Inc.
▶ Home values may be related to such factors as home and property size, location, number of bedrooms, and number of bathrooms
▶Dependent variable: ▶ home values
▶Independent variables: ▶ home and property size
▶ location
▶ number of bedrooms
▶ number of bathrooms
Associative Forecasting
4 - 47© 2014 Pearson Education, Inc.
Forecasting an outcome based on predictor variables using the least squares technique
y = a + bx
where y= computed value of the variable to be predicted (dependent variable)x= the independent variable though to predict the value of the dependent variable
22 xnx
yxnxyb
xbya
Associative Forecasting
4 - 48© 2014 Pearson Education, Inc.
▶Coefficient of determination▶How strong is the linear relationship between the variables?
▶ 0 r2 1
▶The larger the r2 value is, the more valid the linear causal model would be.
▶Physical and medical sciences, r2 0.60
▶Social sciences, r2 0.25
2222
2
2
ynyxnx
yxnxyr
Associative Forecasting
4 - 49© 2014 Pearson Education, Inc.
Example 4.7: Chicken Palace in Spokane, WA, periodically offers carryout five-piece chicken dinners at special prices. The following table summarizes the observations over the past six weeks:
(1) Plot the price against the sales to see if an associative model is appropriate
Price ($) # of dinners sold
2.70 760
3.50 510
2.00 980
4.20 250
3.10 620
4.05 480
0200400600800
10001200
0 1 2 3 4 5
Price ($)
Sal
es (
no
. o
f d
inn
ers)
There is a negative linear relationship between the price and the sales of the chicken dinner. As such, an associative model will be appropriate.
4 - 50© 2014 Pearson Education, Inc.
(2) Develop the associative model to predict future sales of the chicken dinner
x y x2 xy y2
2.70 760
3.50 510
2.00 980
4.20 250
3.10 620
4.05 480
19.55 3600Sum
258.36
55.19x
6006
600,3y
805.289
258.361925.67
600258.36713,102
22
xnx
yxnxyb
185.544,1258.3805.289600 xbya
The associative model is y = 1,544.185 – 289.805x
7.29
12.25
4
17.64
9.61
16.4025
67.1925
2052
1785
1960
1050
1922
1944
10,713
577,600
260,100
960,400
62,500
384,400
230,400
2,475,400
4 - 51© 2014 Pearson Education, Inc.
(3) Compute the coefficient of determination to assess the validity of model
93.0
6006400,475,2258.361925.67
600258.36713,1022
2
2222
2
2
ynyxnx
yxnxyr
As 0.93 > 0.25 (for social sciences), the associative model is valid.
x y x2 xy y2
2.70 760
3.50 510
2.00 980
4.20 250
3.10 620
4.05 480
19.55 3600Sum
7.29
12.25
4
17.64
9.61
16.4025
67.1925
2052
1785
1960
1050
1922
1944
10,713
577,600
260,100
960,400
62,500
384,400
230,400
2,475,400
4 - 52© 2014 Pearson Education, Inc.
(4) how many dinners can Chicken Palace expect to sell per week if the price is set at $3.90 each?
Y = 1,544.185 – 289.805X = 1,544.185 – 289.805(3.9) = 413.95 or about 414
4 - 53© 2014 Pearson Education, Inc.
EX4 in classHealthy Hamburgers has a chain of 6 stores in northern Illinois. Sales figures and profits for the stores are given in the following table. 1)Obtain the associative model and predict profit for a store assuming sales of $10 million.2)Determine the correlation coefficient and interpret it.
Sales, x(in $ millions)
Profits, y(in $ millions)
$6 $0.1316 0.2412 0.2020 0.4415 0.347 0.17
4 - 54© 2014 Pearson Education, Inc.
ACCURACY OF FORECAST
4 - 55© 2014 Pearson Education, Inc.
The objective is to obtain the most accurate forecast no matter the technique
We generally do this by selecting the model that gives us the lowest forecast error
Forecast error = Actual demand – Forecast value
= At – Ft
Accuracy of Forecasts
4 - 56© 2014 Pearson Education, Inc.
Common Measures of Error
Mean Absolute Deviation (MAD)
Mean Forecast Error (Bias)
n
Forecast - Actual
Bias
where n is the number of forecast errors involved
25
27
30
32
28
29
2
3
2
-4
1
6
10
2
3
2
4
1
6
18
Example 4.8: Given the demand data that follow, prepare a forecast by using Naïve approach, simple moving average approach and exponential approach. Compute Bias and MAD.
MAD = = 18/6 = 3 n
n
1 t
tt FA 67.1
6
10Bias
n
1 t
n
FA tt
YearTime Period
(t)Sales (At)
Forecast (Ft) et = At - Ft |et| = |At - Ft|
2007 1 25
2008 2 27
2009 3 30
2010 4 32
2011 5 28
2012 6 29
2013 7 35Sum
(1) Naïve approach
© 2014 Pearson Education, Inc. 4 - 58
27.33
29.67
30
29.67
4.67
-1.67
-1
5.33
7.33
4.67
1.67
1
5.33
12.67
(2) Simple Moving Average (3-year SMA)
MAD = = 12.67/4 = 3.17 n
n
1 t
tt FA 83.1
4
33.7Bias
n
1 t
n
FA tt
YearTime Period
(t)Sales (At)
Forecast (Ft) et = At - Ft |et| = |At - Ft|
2007 1 25
2008 2 27
2009 3 30
2010 4 32
2011 5 28
2012 6 29
2013 7 35Sum
© 2014 Pearson Education, Inc. 4 - 59
25
25
26.2
28.48
30.59
29.04
29
0
2
3.8
3.52
-2.59
-0.04
6
12.69
0
2
3.8
3.52
2.59
0.04
6
17.95
(3) Exponential Smoothing ( = 0.6, F1 = 25, F2 = 25)
MAD = = 17.95/7 = 2.56 n
n
1 t
tt FA 81.1
7
69.12Bias
n
1 t
n
FA tt
YearTime
Period (t)Sales (At)
Forecast (Ft) et = At - Ft |et| = |At - Ft|
2007 1 25
2008 2 27
2009 3 30
2010 4 32
2011 5 28
2012 6 29
2013 7 35Sum
4 - 60© 2014 Pearson Education, Inc.
▶ If Bias is used Naive approach
▶ If MAD is used Exponential smoothing approach
(4) Comparison
Naïve
Approach SMAExponential Smoothing
Bias 1.67 1.83 1.81
MAD 3 3.17 2.56
4 - 61© 2014 Pearson Education, Inc.
EX5 in classData collected on the annual promotional expenditure (in millions of dollars) of Fairface Cosmetics Company over the past seven years are shown in the following table:
1) Use the 2-year SMA method to develop a forecast of the promotional expenditure in each year up to 2013.
2) Repeat (1) by using the 3-year WMA method with weights of 0.5, 0.25, and 0.25.
3) Repeat (1) by using the ES method with α = 0.5 and the forecast for 2007 of $6 million.
4) Which of the three methods is most reliable on the basis of MAD?
5) Based on the findings in (4) above, provide a forecast of the promotional expenditure in 2014.
Year 2007 2008 2009 2010 2011 2012 2013Expenditure 10 8 7 9 12 14 11