Microeconometrics Using Stata. A. COLIN CAMERON. Department of Economics. University of California. Davis, CA. PRAVIN K. TRIVEDI. Department of Economics. Indiana University. Bloomington, IN. A Stata Press Publication. StataCorp LP. College Station,
Njazi Bytyqi, University of Prishtina, Kosovo. Sali Aliu, University of Prishtina, Kosovo. Jehona Shkodra, University of Prishtina, Kosovo. Chapter 60. Application of CRM 2.0 in Spanish Public Administration: Identifying Practical Results. 1228. Dani
You miss 100% of the shots you don’t take. Wayne Gretzky
Idea Transcript
Business Forecasting With
ForecastX ™ Sixth Edition
J. Holton Wilson
Barry Keating
Central Michigan University
University of Notre Dame
John Gait Solutions, Inc. Chicago
Boston Burr Ridge, IL Dubuque, IA Madison, Wl New York San Francisco St. Louis Bangkok Bogota Caracas Kuala Lumpur Lisbon London Madrid Mexico City Milan Montreal New Delhi Santiago Seoul Singapore Sydney Taipei Toronto
Contents Analog Forecasts 22 New Product and Penetration Curves for VCR Sales 23 Test Marketing 24 Product Clinics 24 Type of Product Affects New-Product Forecasting 25 The Bass Model for New-Product Forecasting 25 Forecasting Sales for New Products That Have Short Product Life Cycles 2 7
Chapter 1
Introduction to Business Forecasting 1 Introduction 1 Comments from the Field 2 Quantitative Forecasting Has Become Widely Accepted 2 Forecasting in Business Today 3 Krispy Kreme 5 BellAtlantic 5 Columbia Gas 6 Segix Italia 6 Pharmaceuticals in Singapore 6 Fiat Auto 7 Brake Parts, Inc. 7 Some Global Forecasting Issues: Examples from Ocean Spray Cranberries 7
Forecasting in the Public and Not-for-Profit Sectors 8 Forecasting and Supply Chain Management 10 Collaborative Forecasting 12 Computer Use and Quantitative Forecasting Qualitative or Subjective Forecasting Methods 16 Sales Force Composites 16 Surveys of Customers and the General Population 18 Jury of Executive Opinion 18 The Delphi Method 18 Some Advantages and Disadvantages of Subjective Methods 19
New-Product Forecasting
20
Using Marketing Research to Aid New-Product Forecasting 20 The Product Life Cycle Concept Aids in New-Product Forecasting 21 viii
Two Simple Naive Models 30 Evaluating Forecasts 34 Using Multiple Forecasts 36 Sources of Data 37 Forecasting Total Houses Sold 37 Overview of the Text 39 Comments from the Field 41 Integrative Case: Forecasting Sales of The Gap 41 Comments from the Field 47 John Gait Partial Customer List 48 An Introduction to ForecastX 7.0 49 15
Forecasting with the ForecastX Wizard^™ 49 Using the Five Main Tabs on the Opening ForecastX Screen 49
Suggested Readings and Web Sites Exercises 53
Chapter 2
The Forecast Process, Data Considerations, and Model Selection 56 Introduction 56 The Forecast Process 56 Trend, Seasonal, and Cyclical Data Patterns 59
52
Contents ix
Data Patterns and Model Selection 62 A Statistical Review 64
Forecasting Jewelry Sales and Houses Sold with Exponential Smoothing 143
Descriptive Statistics 64 The Normal Distribution 69 The Student's t-Distribution 71 From Sample to Population: Statistical Inference 74 Hypothesis Testing 76 Correlation 81
Correlograms: Another Method of Data Exploration 84 Total Houses Sold: Exploratory Data Analysis and Model Selection 87 Business Forecasting: A Process, Not an Application 89 Integrative Case: The Gap 89 Comments from the Field 92 Using ForecastX™ to Find Autocorrelation Functions 93 Suggested Readings 95 Exercises 96
Chapter 3 Moving Averages and Exponential Smoothing 101 Moving Averages 101 Simple Exponential Smoothing ,107 Holt's Exponential Smoothing 112 Winters' Exponential Smoothing 118 The Seasonal Indices
120
Adaptive-Response-Rate Single Exponential Smoothing 121 Using Single, Holt's, or ADRES Smoothing to Forecast a Seasonal Data Series 124 New-Product Forecasting (Growth Curve Fitting) 125 Gompertz Curve 129 Logistics Curve 133 Bass Model 135 The Bass Model in Action
Event Modeling
13 9
136
Jewelry Sales 143 Houses Sold 145
Summary 146 Integrative Case: The Gap 147 Using ForecastX™ to Make Exponential Smoothing Forecasts 149 Suggested Readings 151 Exercises 152
Chapter 4 Introduction to Forecasting with Regression Methods 160 The Bivariate Regression Model 160 Visualization of Data: An Important Step in Regression Analysis 161 A Process for Regression Forecasting 164 Forecasting with a Simple Linear Trend 165 Using a Causal Regression Model to Forecast 171 A Jewelry Sales Forecast Based on Disposable Personal Income 173 Statistical Evaluation of Regression Models 178 Basic Diagnostic Checks for Evaluating Regression Results 178
Using the Standard Error of the Estimate Serial Correlation 185 Heteroscedasticity 190 Cross-Sectional Forecasting 191 Forecasting Total Houses Sold with Two Bivariate Regression Models 193 Comments from the Field 200 Integrative Case: The Gap 200 Comments from the Field 204 Using ForecastX™ to Make Regression Forecasts 205 Further Comments on Regression Models 210 Suggested Readings 213 Exercises 214
184
x
Contents
Chapter 5 Forecasting with Multiple Regression 225
The Cycle Factorfor Private Housing Starts 311
The Time-Series Decomposition Forecast 315
The Multiple-Regression Model 225 Selecting Independent Variables 226 Forecasting with a Multiple-Regression Model 227 The Regression Plane
233
Statistical Evaluation of Multiple-Regression Models 235 Three Quick Checks in Evaluating MultipleRegression Models 235 Multicollinearity 240 The Demand for Nurses 241 Serial Correlation: A Second Look 242
Serial Correlation and the Omitted-Variable Problem
Forecasting Shoe Store Sales by Using Time-Series Decomposition 316 Forecasting Total Houses Sold by Using Time-Series Decomposition 319 Forecasting Winter Daily Natural Gas Demand at Vermont Gas Systems 321
Integrative Case: The Gap 321 Using ForecastX™ to Make Time-Series Decomposition Forecasts 325 Suggested Readings 327 Exercises 327 Appendix: Components of the Composite Indices 339
244
Alternative-Variable Selection Criteria 246
Accounting for Seasonality in a Multiple-Regression Model 248 Extensions of the Multiple-Regression Model 260 Advice on Using Multiple Regression in Forecasting 262 Forecasting Jewelry Sales with Multiple Regression 267 Forecasting Consumer Products 275 Integrative Case: The Gap 278 Using ForecastX™ to Make Multiple-Regression Forecasts 282 Suggested Readings 284 Exercises 284
Chapter 6 Time-Series Decomposition
298
The Basic Time-Series Decomposition Model 298 Deseasonalizing the Data and Finding Seasonal Indices 301 Finding the Long-Term Trend 308 Measuring the Cyclical Component 308 Overview of Business Cycles 309 Business Cycle Indicators 310
Chapter 7 ARIMA (Box-Jenkins)-Type Forecasting Models 343 Introduction 343 The Philosophy of Box-Jenkins 344 Moving-Average Models 346 Autoregressive Models 3 51 Mixed Autoregressive and Moving-Average Models 356 Stationarity 357 The Box-Jenkins Identification Process 361 Comments from the Field: An Overview of INTELSAT Forecasting 366 ARIMA: A Set of Numerical Examples 366 Example 1 Example 2 Example 3 Example 4
366 367 370 372
Forecasting Seasonal Time Series 379 Total Houses Sold 379 Intelligent Transportation Systems 383 Integrative Case: Forecasting Sales of The Gap 385 Using ForecastX™ to Make ARIMA (Box-Jenkins) Forecasts 390
Contents xi
Suggested Readings 392 Exercises 393 Appendix: Critical Values of Chi-Square
401
Chapter 8
Combining Forecast Results
402
Introduction 402 Bias 404 An Example 404 What Kinds of Forecasts Can Be Combined? 408 Considerations in Choosing the Weights for Combined Forecasts 409 Three Techniques for Selecting Weights When Combining Forecasts 413 Justice Is Forecast 415 An Application of the Regression Method for Combining Forecasts 416
Forecasting Total Houses Sold with a Combined Forecast 419 Comments from the Field: Combining Forecasts Can Improve Results 425 Integrative Case: Forecasting The Gap Sales Data with a Combination Model 426 Using ForecastX™ to Combine Forecasts 430 Suggested Readings 432 Exercises 433
Chapter 9
Data Mining 439 Introduction 439 Data Mining 440 Comments from the Field 443 The Tools of Data Mining 443 Business Forecasting and Data Mining
444
A Data Mining Example: k-Nearest-Neighbor 445 Comments from the Field: Cognos 449 A Business Data Mining Example: k-Nearest-Neighbor 449 Classification Trees: A Second Classification Technique 457 A Business Data Mining Example: Classification Trees 461
Naive Bayes: A Third Classification Technique 464 Comments from the Field 472 Regression: A Fourth Classification Technique 472 Comments from the Field: Fool's Gold 478 Summary 479 Suggested Readings 479 Exercises 479
Chapter 10
Forecast Implementation
482
Keys to Obtaining Better Forecasts The Forecast Process 485
482
Step 1. Specify Objectives 486 Step 2. Determine What to Forecast 487 Step 3. Identify Time Dimensions 487 Step 4. Data Considerations 487 How to Evaluate and Improve a Forecasting Process 488 Step 5. Model Selection 488 Step 6. Model Evaluation 489 Step 7. Forecast Preparation 489 Step 8. Forecast Presentation 490 Step 9. Tracking Results 490
Choosing the Right Forecasting Techniques 491 Sales Force Composite (SFC) 491 Customer Surveys (CS) 493 Jury of Executive Opinion (JEO) 493 Delphi Method 493 Naive 494 Moving Averages 494 Simple Exponential Smoothing (SES) 494 Adaptive-Response-Rate Single Exponential Smoothing (ADRES) 494 Holt s Exponential Smoothing (HES) 495 Winters 'Exponential Smoothing (WES) 495 Regression-Based Trend Models 495 Regression-Based Trend Models with Seasonally 495 Comments from the Field 496 Regression Models with Causality 496
xii Contents Comments from the Field 497 Time-Series Decomposition (TSD) 497 ARIMA 498
Special Forecasting Considerations
498
Event Modeling 498 Combining Forecasts 499 New-Product Forecasting (NPF) 499 Data Mining 500 Comments from the Field 501
Summary 502 Using ProCast™ in ForecastX™ to Make Forecasts 503 Suggested Readings 505 Exercises 506 Index