assignment 8 &9

Chapter 9

Forecasting

1

Chapter Objectives
Be able to:
Differentiate between demand forecasts, supply forecasts, and price forecasts.
Describe the four laws of forecasting.
Select the most appropriate type of forecasting approach, given different forecasting situations.
Describe four qualitative forecasting techniques – market surveys, panel consensus forecasting, the Delphi method, and the life cycle analogy – and explain when they should be used.
Apply a variety of time series forecasting models, including moving average, exponential smoothing, and linear regression models.
Develop causal forecasting models using linear regression and multiple regression.
Calculate measures of forecasting accuracy and interpret the results.
Describe the benefits of using computer-based forecasting packages.
Explain what collaborative planning, forecasting, and replenishment (CPFR) is and how it helps supply chain partners synchronize their plans and actions.

Copyright © 2016 Pearson Education, Inc.
9-2

9

Introduction
Forecast – An estimate of the future level of some variable.

Forecasting is used to determine:
Long-term capacity needs
Yearly business plans
Shorter-term operations and supply chain activities

Copyright © 2016 Pearson Education, Inc.
9-3

9

Forecast Types
Demand
Overall market demand
Firm-Level demand
Supply
Number of current producers and suppliers
Projected aggregate supply levels
Technological and political trends
Price
Forecast prices for key materials and services

Copyright © 2016 Pearson Education, Inc.
9-4

9

s of Forecasting
1: Forecasts Are Almost Always Wrong (But They Are Still Useful).
2: Forecasts for the Near Term Tend To Be More Accurate.
3: Forecasts for Groups of Products or Services Tend to Be More Accurate.
4: Forecasts Are No Substitute For Calculated Values.
Copyright © 2016 Pearson Education, Inc.
9-5

9

Selecting a Forecasting Method
Qualitative forecasting techniques – Forecasting techniques based on intuition or informed opinion.
Used when data are scarce, not available, or irrelevant.

Quantitative forecasting models – Forecasting models that use measurable or historical data to generate forecasts.
Time series and causal models
Copyright © 2016 Pearson Education, Inc.
9-6

9

Selecting a Forecasting Method

Figure 9.2
Copyright © 2016 Pearson Education, Inc.
9-7

9

Qualitative Forecasting Methods
Market surveys
Panel consensus forecasting
Delphi method
Life-cycle analogy method
Build-up forecasts

Copyright © 2016 Pearson Education, Inc.
9-8

9

Time Series Forecasting Methods
Time series forecasting models – A quantitative forecasting model that uses a time series to develop forecasts.
Time series – A series of observations arranged in chronological order

Figure 9.3
Copyright © 2016 Pearson Education, Inc.
9-9

9

9

Time Series Forecasting Methods
Demand Patterns
Randomness – Unpredictable movement from one time period to the next.

Trend – Long-term movement up or down in a time series.

Seasonality – A repeated pattern of spikes or drops in a time series associated with certain times of the year.
Copyright © 2016 Pearson Education, Inc.
9-10

9

Time Series Forecasting Methods

Figure 9.4
Time Series Showing Randomness, a Downward Trend, and Seasonality (Higher Demand in the Winter Months
Copyright © 2016 Pearson Education, Inc.
9-11

9

Time Series Forecasting Methods
Last Period Model – The simplest time series model which uses demand for the current period as a forecast for the next period.

Ft+1 = Dt

where Ft+1= forecast for the next period, t+1
and Dt = demand for the current period, t
Copyright © 2016 Pearson Education, Inc.
9-12

9

Time Series Forecasting Methods

Table 9.3
Figure 9.5
Last Period
Copyright © 2016 Pearson Education, Inc.
9-13

9

Time Series Forecasting Methods
Moving Average Model – A time series forecasting model that derives a forecast by taking an average of recent demand values.

Copyright © 2016 Pearson Education, Inc.
9-14

9

Time Series Forecasting Methods

Two-Period and Four-Period Moving Average Forecasts
Table 9.4
Example: Week 16 = (101 + 109)/2 = 105
Copyright © 2016 Pearson Education, Inc.
9-15

9

Time Series Forecasting Methods

Two-Period and Four-Period Moving Average Forecasts
Figure 9.6

Copyright © 2016 Pearson Education, Inc.
9-16

9

Time Series Forecasting Methods
Weighted Moving Average Model – A form of the moving average model that allows the actual weights applied to past observations to differ.

Copyright © 2016 Pearson Education, Inc.
9-17

9

Example 9.1 – Flavio’s Pizza
Flavio’s Pizza has recorded the following demand history for each Friday night for the past five weeks.
Develop forecasts for week 6 using a two-period moving average and a 3-period weighted moving average using the following demands and weights (0.4, 0.35, and 0.25, starting with the most recent observation.

The three-period weighted moving average forecast would be:
The two-period moving average forecast would be:

Copyright © 2016 Pearson Education, Inc.
9-18

9

Time Series Forecasting Methods
Exponential Smoothing Model – A form of the moving average model in which the forecast for the next period is calculated as the weighted average of the current period’s actual value and forecast.

Copyright © 2016 Pearson Education, Inc.
9-19

9

Time Series Forecasting Methods
Exponential Smoothing Model
The general rule for determining the a value:
The greater the randomness in the time series data, the lower the a value should be.
The less randomness in the time series data, the higher the a value should be.
Copyright © 2016 Pearson Education, Inc.
9-20

9

Example 9.2 – Exponential Smoothing

Using the following time series data, calculate an exponential smoothing forecast for periods 2 through 20 using a smoothing constant value of 0.8

Figure 9.8
Copyright © 2016 Pearson Education, Inc.
9-21

9

Example 9.2 – Exponential Smoothing

The detailed calculations for F2 through F7 are:
Figure 9.9
Copyright © 2016 Pearson Education, Inc.
9-22

9

Example 9.2 – Exponential Smoothing

Figure 9.9
Exponential Smoothing Forecast (a = 0.8)
Copyright © 2016 Pearson Education, Inc.
9-23

9

Time Series Forecasting Methods
Adjusted Exponential Smoothing Model – An expanded version of the exponential smoothing model that includes a trend adjustment factor.

AFt+1 = Ft+1 +Tt+1

where AFt+1 = adjusted forecast for the next period
Ft+1 = unadjusted forecast for the next period = Dt + (1 – ) Ft
Tt+1 = trend factor for the next period = (Ft+1 – Ft) + (1 – )Tt
Tt = trend factor for the current period
 = smoothing constant for the trend adjustment factor

Copyright © 2016 Pearson Education, Inc.
9-24

9

Time Series Forecasting Models
Linear Regression – A statistical techniques that expresses a forecast variable as a linear function of some independent variable.
Can be used to develop time series and causal forecasting models.

Copyright © 2016 Pearson Education, Inc.
9-25

9

Time Series Forecasting Models
Linear Regression Calculations for the y intercept (a) and slope (b)

Copyright © 2016 Pearson Education, Inc.
9-26

9

Example 9.3 – Clem’s Competition Clutches
Mike Clem, owner of Clem’s
Competition Clutches, designs
and manufacturers heavy-duty car
clutches for use in drag racing.
In his first 10 months of business,
Mike has experienced the demand
shown in Table 9.7 and Figure 9.11.
Use Linear Regression to forecast
demand for months 11, 12, and 13.

Table 9.7
Copyright © 2016 Pearson Education, Inc.
9-27

9

Example 9.3 – Clem’s Competition Clutches

Ten-Month Time Series of Demand
Figure 9.11
Copyright © 2016 Pearson Education, Inc.
9-28

9

Example 9.3 – Clem’s Competition Clutches

Copyright © 2016 Pearson Education, Inc.
9-29

9

Example 9.3 – Clem’s Competition Clutches

Figure 9.12
Copyright © 2016 Pearson Education, Inc.
9-30

9

Time Series Forecasting Models
Seasonality – Repeated patterns or drops in a time series associated with certain times of the year.

Table 9.8

Examples of Products and Service That Experience Seasonality
Copyright © 2016 Pearson Education, Inc.
9-31

9

Time Series Forecasting Models
Seasonal Adjustments Four-Step Procedure:
For each of the demand values in the time series, calculate the corresponding forecast using the unadjusted forecast model.
For each demand value, calculate (Demand/Forecast). If the ratio is less than 1, then the forecast model overforecasted; if it is greater than 1, then the model underforecasted.
If the time series covers multiple years, take the average (Demand/Forecast) for corresponding months or quarters to derive the seasonal index. Otherwise use (Demand/Forecast) calculated in Step 2 as the seasonal index.
Multiply the unadjusted forecast by the seasonal index to get the seasonally adjusted forecast value.
Copyright © 2016 Pearson Education, Inc.
9-32

9

Example 9.4 – Seasonal Adjustments

Figure 9.15

Based on the results of the regression model, develop a seasonal index for each month and reforecast months 1 through 24 (January 2016 – December 2017) using the seasonal indices.
Copyright © 2016 Pearson Education, Inc.
9-33

9

Example 9.4 – Seasonal Adjustments

Copyright © 2016 Pearson Education, Inc.
9-34

9

Example 9.4 – Seasonal Adjustments

Copyright © 2016 Pearson Education, Inc.
9-35

9

Example 9.4 – Seasonal Adjustments
Calculate the (Demand/Forecast) for each of the time periods:
January 2012: (Demand/Forecast) = 51/106.9 = 0.477

January 2013: (Demand/Forecast) = 112/205.6 = 0.545

Calculate the monthly seasonal indices:
Monthly seasonal index, January = (.477 + .545)/2 = .511
Calculate the seasonally adjusted forecasts
Seasonally adjusted forecast = unadjusted forecast x seasonal index
January 2012: 106.9 x .511 = 54.63
January 2013: 205.6 x .511 = 105.06
Copyright © 2016 Pearson Education, Inc.
9-36

9

Example 9.4 – Seasonal Adjustments
Copyright © 2016 Pearson Education, Inc.
9-37

Figure 9.16

9

Example 9.4 – Seasonal Adjustments

Plot of Seasonally Adjusted Regression Forecast against a Time Series Showing Seasonality
Figure 9.16
Copyright © 2016 Pearson Education, Inc.
9-38

9

Causal Forecasting Models
Causal Forecasting Models – A class of quantitative forecasting models in which the forecast is modeled as a function of something other than time.
Linear Regression
Multiple Regression

Copyright © 2016 Pearson Education, Inc.
9-39

9

Causal Forecasting Models
Linear Regression
Copyright © 2016 Pearson Education, Inc.
9-40

Forecasted home sales at 6% mortgage rate:
1,488.58 – 9,846.94(6%) = 898 home sales
Forecasted home sales at 8% mortgage rate:
1,488.58 – 9,846.94(8%) = 701 home sales
Forecasted home sales =
1,488.58 – 9,846.94(mortgage rate)

9

Causal Forecasting Models
Multiple Regression – A generalized form of linear regression that allows for more than one independent variable.

Copyright © 2016 Pearson Education, Inc.
9-41

9

Example 9.6 – Lance’s BBQ
Lance’s BBQ Catering Service is a favorite of sports teams in the Raleigh, North Carolina area.
By counting and weighing the guests arriving at a party, they captured the following data:
Copyright © 2016 Pearson Education, Inc.
9-42

9

Example 9.6 – Lance’s BBQ
Lance has a party coming up and is expecting 60 guests each with an average weight of around 240 lbs.

Use multiple regression to estimate how much barbecue these guests will eat based on the number of guests and average weight.
Copyright © 2016 Pearson Education, Inc.
9-43

9

Example 9.6 – Lance’s BBQ
Copyright © 2016 Pearson Education, Inc.
9-44
Intercept term = 12.52
Slope coefficient for number of guests = 0.15
Slope coefficient for average weight = 0.15

Barbecue eaten (lbs.) = 12.52 + 0.15 (no. of guests) + 0.15 (average weight)

If expecting 60 guests with average weight of 240 lbs.

Barbecue eaten (lbs.) = 12.52 + 0.15 (60) + 0.15 (240)

= 57.52 lbs. of barbecue

9

Measures of Forecast Accuracy
Measures of Forecast Accuracy are used to assess how well an individual model is performing or to compare multiple forecast models to one another.
Copyright © 2016 Pearson Education, Inc.
9-45

9

Measures of Forecast Accuracy
Forecast error for period i (FEi) =

Mean forecast error (MFE) =

Mean absolute deviation (MAD) =

Copyright © 2016 Pearson Education, Inc.
9-46

9

Measures of Forecast Accuracy
Mean absolute percentage error (MAPE) =

Tracking Signal =

Copyright © 2016 Pearson Education, Inc.
9-47

9

Example 9.7 – Wolf State University
Andi Irby, director of advising is trying to decide which of two forecasting models does a better job at predicting walk-in demand for student advising.
Copyright © 2016 Pearson Education, Inc.
9-48

9

Example 9.7 – Wolf State University
Table 9.11

Demand and Forecast Results for Walk-In Advising
at Wolf State University
Copyright © 2016 Pearson Education, Inc.
9-49

9

Example 9.7 – Wolf State University

Copyright © 2016 Pearson Education, Inc.
9-50

9

Example 9.7 – Wolf State University
Model 2 has the MFE value closer to 0, it appears to be the least biased.
On average, Model 2 overforecasted by 0.20 walk-ins, while Model 1 overforecasted by 0.70.
Model 2 has the lower MAD and MAPE values.
Model 2 is the superior model.
Andi also develop a tracking signal for the first 20 weeks. The details are shown on the following slide.

Copyright © 2016 Pearson Education, Inc.
9-51

9

Example 9.7 – Wolf State University

Copyright © 2016 Pearson Education, Inc.
9-52

9

Computer-Based Forecasting Packages
Computer-based forecasting packages are used to develop, evaluate and change forecasting models as needed.
With enough demand history, a package could quickly evaluate alternative forecasting methods for each item and select the model that best fits the past data.
Some packages can use MFE, MAD, and MAPE or tracking signal criteria to flag a poor forecasting model and kick off a search for a new one. Others can develop multiples forecasts for a single item.

Copyright © 2016 Pearson Education, Inc.
9-53

9

Collaborative Planning,
Forecasting, and Replenishment (CPFR)
CPFR – A set of business processes, backed up by information technology, in which supply chain partners agree to mutual business objectives and measures, develop joint sales and operational plans, and collaborate to generate and update sales forecasts and replenishment plans.
Copyright © 2016 Pearson Education, Inc.
9-54

9

Forecasting Case Study
Top-Slice Drivers
Copyright © 2016 Pearson Education, Inc.
9-55

55

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher.
Printed in the United States of America.
Copyright © 2016 Pearson Education, Inc.
9-56

56

n
D
F
n
i
i
t
t
å
=
–
+
+
=
1
1
1