Forecasting Methods and Applications

by Makridakis, Spyros G.; Wheelwright, Steven C.; Hyndman, Rob J.

• ISBN13: 9780471532330
• ISBN10: 0471532339
• Edition: 3rd
• Format: Hardcover
• Copyright: 1997-12-29
• Publisher: Wiley

Summary

Can You Predict the Future by Looking at the Past? Since accurate forecasting requires more than just inserting historical data into a model, Forecasting: Methods and Applications, 3/e, adopts a managerial, business orientation. Integrated throughout this text is the innovative idea that explaining the past is not adequate for predicting the future. Inside, you will find the latest techniques used by managers in business today, discover the importance of forecasting and learn how it's accomplished. And you'll develop the necessary skills to meet the increased demand for thoughtful and realistic forecasts. New features in the third edition include: * An emphasis placed on the practical uses of forecasting. * All data sets used in this book are available on the Internet. * Comprehensive coverage provided on both quantitative and qualitative forecasting techniques. * Includes many new developments in forecasting methodology and practice.

Author Biography

Spyros Makridakis is a professor of the University of Nicosia UNIC where he is the Director of the Institute for the Future and an Emeritus Professor of Decision Sciences at INSEAD. Steven C. Wheelwright is the author of Forecasting: Methods and Applications, 3rd Edition, published by Wiley.

Table of Contents

1 THE FORECASTING PERSPECTIVE	1	(19)
1/1 Why forecast?	2	(4)
1/2 An overview of forecasting techniques	6	(7)
1/2/1 Explanatory versus time series forecasting	10	(2)
1/2/2 Qualitative forecasting	12	(1)
1/3 The basic steps in a forecasting task	13	(4)
References and selected bibliography	17	(2)
Exercises	19	(1)
2 / BASIC FORECASTING TOOLS	20	(61)
2/1 Time series and cross-sectional data	21	(2)
2/2 Graphical summaries	23	(5)
2/2/1 Time plots and time series patterns	24	(2)
2/2/2 Seasonal plots	26	(1)
2/2/3 Scatterplots	27	(1)
2/3 Numerical summaries	28	(13)
2/3/1 Univariate statistics	29	(5)
2/3/2 Bivariate statistics	34	(4)
2/3/3 Autocorrelation	38	(3)
2/4 Measuring forecast accuracy	41	(11)
2/4/1 Standard statistical measures	42	(3)
2/4/2 Out-of-sample accuracy measurement	45	(1)
2/4/3 Comparing forecast methods	46	(2)
2/4/4 Theil's U-statistic	48	(2)
2/4/5 ACF of forecast error	50	(2)
2/5 Prediction intervals	52	(2)
2/6 Least squares estimates	54	(9)
2/6/1 Discovering and describing relationships	59	(4)
2/7 Transformations and adjustments	63	(8)
2/7/1 Mathematical transformations	63	(4)
2/7/2 Calendar adjustments	67	(3)
2/7/3 Adjustments for inflation and population changes	70	(1)
Appendices	71	(3)
2-A Notation for quantitative forecasting	71	(1)
2-B Summation sign XXX	72	(2)
References and selected bibliography	74	(2)
Exercises	76	(5)
3 / TIME SERIES DECOMPOSITION	81	(54)
3/1 Principles of decomposition	84	(5)
3/1/1 Decomposition models	84	(3)
3/1/2 Decomposition graphics	87	(1)
3/1/3 Seasonal adjustment	88	(1)
3/2 Moving averages	89	(12)
3/2/1 Simple moving averages	89	(5)
3/2/2 Centered moving averages	94	(4)
3/2/3 Double moving averages	98	(1)
3/2/4 Weighted moving averages	98	(3)
3/3 Local regression smoothing	101	(5)
3/3/1 Loess	104	(2)
3/4 Classical decomposition	106	(7)
3/4/1 Additive decomposition	107	(2)
3/4/2 Multiplicative decomposition	109	(3)
3/4/3 Variations on classical decomposition	112	(1)
3/5 Census Bureau methods	113	(8)
3/5/1 First iteration	114	(4)
3/5/2 Later iterations	118	(1)
3/5/3 Extensions to X-12-ARIMA	119	(2)
3/6 STL decomposition	121	(4)
3/6/1 Inner loop	122	(1)
3/6/2 Outer loop	123	(1)
3/6/3 Choosing the STL parameters	124	(1)
3/6/4 Comparing STL with X-12-ARIMA	124	(1)
3/7 Forecasting and decomposition	125	(2)
References and selected bibliography	127	(3)
Exercises	130	(5)
4 / EXPONENTIAL SMOOTHING METHODS	135	(50)
4/1 The forecasting scenario	138	(3)
4/2 Averaging methods	141	(6)
4/2/1 The mean	141	(1)
4/2/2 Moving averages	142	(5)
4/3 Exponential smoothing methods	147	(24)
4/3/1 Single exponential smoothing	147	(8)
4/3/2 Single exponential smoothing: an adaptive approach	155	(3)
4/3/3 Hot's linear method	158	(3)
4/3/4 Holt-Winters' trend and seasonality method	161	(8)
4/3/5 Exponential smoothing: Pegels' classification	169	(2)
4/4 A comparison of methods	171	(3)
4/5 General aspects of smoothing methods	174	(5)
4/5/1 Initialization	174	(2)
4/5/2 Optimization	176	(1)
4/5/3 Prediction intervals	177	(2)
References and selected bibliography	179	(2)
Exercises	181	(4)
5 / SIMPLE REGRESSION	185	(55)
5/1 Regression methods	186	(1)
5/2 Simple regression	187	(21)
5/2/1 Least squares estimation	188	(5)
5/2/2 The correlation coefficient	193	(3)
5/2/3 Cautions in using correlation	196	(2)
5/2/4 Simple regression and the correlation coefficient	198	(5)
5/2/5 Residuals, outliers, and influential observations	203	(5)
5/2/6 Correlation and causation	208	(1)
5/3 Inference and forecasting with simple regression	208	(13)
5/3/1 Regression as statistical modeling	209	(2)
5/3/2 The F-test for overall significance	211	(4)
5/3/3 Confidence intervals for individual coefficients	215	(2)
5/3/4 t-tests for individual coefficients	217	(1)
5/3/5 Forecasting using the simple regression model	218	(3)
5/4 Non-linear relationships	221	(7)
5/4/1 Non-linearity in the parameters	222	(2)
5/4/2 Using logarithms to form linear models	224	(1)
5/4/3 Local regression	224	(4)
Appendixes	228	(2)
5-A Determining the values of a and b	228	(2)
References and selected bibliography	230	(1)
Exercises	231	(9)
6 / MULTIPLE REGRESSION	240	(71)
6/1 Introduction to multiple linear regression	241	(22)
6/1/1 Multiple regression model: theory and practice	248	(2)
6/1/2 Solving for the regression coefficients	250	(1)
6/1/3 Multiple regression and the coefficient of determination	251	(1)
6/1/4 The F-test for overall significance	252	(3)
6/1/5 Individual coefficients: confidence intervals and t-tests	255	(4)
6/1/6 The assumptions behind multiple linear regression models	259	(4)
6/2 Regression with time series	263	(11)
6/2/1 Checking independence of residuals	265	(4)
6/2/2 Time-related explanatory variables	269	(5)
6/3 Selecting variables	274	(13)
6/3/1 The long list	276	(1)
6/3/2 The short list	277	(2)
6/3/3 Best subsets regression	279	(6)
6/3/4 Stepwise regression	285	(2)
6/4 Multicollinearity	287	(4)
6/4/1 Multicollinearity when there are two regressors	289	(1)
6/4/2 Multicollinearity when there are more than two regressors	289	(2)
6/5 Multiple regression and forecasting	291	(8)
6/5/1 Example: cross-sectional regression and forecasting	292	(2)
6/5/2 Example: time series regression and forecasting	294	(4)
6/5/3 Recapitulation	298	(1)
6/6 Econometric models	299	(4)
6/6/1 The basis of econometric modeling	299	(2)
6/6/2 The advantages and drawbacks of econometric methods	301	(2)
Appendixes	303	(2)
6-A The Durbin-Watson statistic	303	(2)
References and selected bibliography	305	(1)
Exercises	306	(5)
7 / THE BOX-JENKINS METHODOLOGY FOR ARIMA MODELS	311	(77)
7/1 Examining correlations in times series data	313	(11)
7/1/1 The autocorrelation function	313	(4)
7/1/2 A white noise model	317	(1)
7/1/3 The sampling distribution of autocorrelations	317	(1)
7/1/4 Portmanteau tests	318	(2)
7/1/5 The partial autocorrelation coefficient	320	(2)
7/1/6 Recognizing seasonality in a time series	322	(1)
7/1/7 Example: Pigs slaughtered	322	(2)
7/2 Examining stationarity of time series data	324	(11)
7/2/1 Removing non-stationarity in a time series	326	(3)
7/2/2 A random walk model	329	(1)
7/2/3 Tests for statationarity	329	(2)
7/2/4 Seasonal differencing	331	(3)
7/2/5 Backshift notion	334	(1)
7/3 ARIMA models for times series data	335	(12)
7/3/1 An autoregressive model of order one	337	(2)
7/3/2 A moving average model of order one	339	(1)
7/3/3 Higher-order autoregressive models	339	(3)
7/3/4 Higher-order moving average models	342	(2)
7/3/5 Mixtures: ARIMA models	344	(1)
7/3/6 Mixtures: ARIMA models	345	(1)
7/3/7 Seasonality and ARIMA models	346	(1)
7/4 Identification	347	(11)
7/4/1 Example 1: A non-seasonal time series	349	(3)
7/4/2 Example 2: A seasonal time series	352	(2)
7/4/3 Example 3: A seasonal time series needing transformation	354	(3)
7/4/4 Recapitulation	357	(1)
7/5 Estimating the parameters	358	(2)
7/6 Identification revisited	360	(4)
7/6/1 Example 1: Internet usage	362	(1)
7/6/2 Example 2: Sales of printing/writing paper	362	(2)
7/7 Diagnostic checking	364	(2)
7/8 Forecasting with ARIMA models	366	(8)
7/8/1 Point forecasts	366	(4)
7/8/2 Out-of-sample forecasting	370	(1)
7/8/3 The effect of differencing on forecasts	371	(1)
7/8/4 ARIMA models used in time series decomposition	372	(1)
7/8/5 Equivalances with exponential smoothing models	373	(1)
References and selected bibliography	374	(3)
Exercises	377	(11)
8 / ADVANCED FORECASTING MODELS	388	(63)
8/1 Regression with ARIMA errors	390	(13)
8/1/1 Modeling procedure	391	(2)
8/1/2 Example: Japanese motor vehicle production	393	(3)
8/1/3 Example: Sales of petroleum and coal products	396	(4)
8/1/4 Forecasting	400	(3)
8/2 Dynamic regression models	403	(15)
8/2/1 Lagged explanatory variables	403	(3)
8/2/2 Koyck model	406	(1)
8/2/3 The basic forms of the dynamic regression model	407	(2)
8/2/4 Selecting the model order	409	(4)
8/2/5 Forecasting	413	(2)
8/2/6 Example: Housing starts	415	(3)
8/3 Intervention analysis	418	(5)
8/3/1 Step-based interventions	419	(2)
8/3/2 Pulse-based interventions	421	(1)
8/3/3 Further reading	422	(1)
8/3/4 Intervention models and forecasting	423	(1)
8/4 Multivariate autoregressive models	423	(6)
8/5 State space models	429	(4)
8/5/1 Some forecasting models in state space form	429	(2)
8/5/2 State space forecasting	431	(2)
8/5/3 The value of state space models	433	(1)
8/6 Non-linear models	433	(2)
8/7 Neural network forecasting	435	(5)
References and selected bibliography	440	(4)
Exercises	444	(7)
9 / FORECASTING THE LONG-TERM	451	(31)
9/1 Cycles versus long-term trends: forecasting copper prices	452	(7)
9/1/1 Forecasting IBM's sales	457	(2)
9/2 Long-term mega economic trends	459	(7)
9/2/1 Cycles of various durations and depths	461	(3)
9/2/2 Implications of extrapolating long-term trends	464	(2)
9/3 Analogies	466	(6)
9/3/1 The Information versus the Industrial Revolution	467	(2)
9/3/2 Five major inventions of the Industrial Revolution and their analogs	469	(3)
9/4 Scenario building	472	(6)
9/4/1 Businesses: gaining and/or maintaining competitive advantages	472	(3)
9/4/2 Jobs, work, and leisure time	475	(1)
9/4/3 Physical versus tele-interactions: extent and speed of acceptance	476	(2)
References and selected bibliography	478	(2)
Exercises	480	(2)
10 / JUDGMENTAL FORECASTING AND ADJUSTMENTS	482	(32)
10/1 The accuracy of judgmental forecasts	483	(9)
10/1/1 The accuracy of forecasts in financial and other markets	484	(6)
10/1/2 Non-investment type forecasts	490	(2)
10/2 The nature of judgmental biases and limitations	492	(11)
10/2/1 Judgmental biases in forecasting	493	(3)
10/2/2 Dealing with judgmental biases	496	(6)
10/2/3 Conventional wisdom	502	(1)
10/3 Combining statistical and judgmental forecasts	503	(5)
10/3/1 Arriving at final forecasts during a budget meeting	503	(5)
10/4 Conclusion	508	(1)
References and selected bibliography	509	(3)
Exercises	512	(2)
11 / THE USE OF FORECASTING METHODS IN PRACTICE	514	(35)
11/1 Surveys among forecasting users	515	(10)
11/1/1 Familiarity and satisfaction with major forecasting methods	516	(4)
11/1/2 The use of different forecasting methods	520	(5)
11/2 Post-sample accuracy: empirical findings	525	(7)
11/3 Factors influencing method selection	532	(5)
11/4 The combination of forecasts	537	(6)
11/4/1 Factors that contribute to making combining work	538	(1)
11/4/2 An example of combining	539	(4)
References and selected bibliography	543	(4)
Exercises	547	(2)
12 / IMPLEMENTING FORECASTING: ITS USES, ADVANTAGES, AND LIMITATIONS	549	(28)
12/1 What can and cannot be predicted	551	(7)
12/1/1 Short-term predictions	553	(1)
12/1/2 Medium-term predictions	554	(3)
12/1/3 Long-term predictions	557	(1)
12/2 Organizational aspects of forecasting	558	(9)
12/2/1 Correcting an organization's forecasting problems	561	(1)
12/2/2 Types of forecasting problems and their solutions	562	(5)
12/3 Extrapolative predictions versus creative insights	567	(4)
12/3/1 Hindsight versus foresight	569	(2)
12/4 Forecasting in the future	571	(4)
12/4/1 Data, information, and forecasts	571	(1)
12/4/2 Collective knowledge, experience, and forecasting	572	(3)
References and selected bibliography	575	(1)
Exercises	576	(1)
APPENDIX I / FORECASTING RESOURCES	577	(12)
1 Forecasting software	578	(5)
1/1 Spreadsheets	578	(1)
1/2 Statistics packages	578	(1)
1/3 Specialty forecasting packages	579	(3)
1/4 Selecting a forecasting package	582	(1)
2 Forecasting associations	583	(2)
3 Forecasting conferences	585	(1)
4 Forecasting journals and newsletters	585	(1)
5 Forecasting on the Internet	586	(2)
References and selected bibliography	588	(1)
APPENDIX II / GLOSSARY OF FORECASTING TERMS	589
APPENDIX III / STATISTICAL TABLES	549	(84)
A: Normal probabilities	620	(1)
B: Critical values for t-statistic	621	(1)
C: Critical values for F-statistic	622	(6)
D: Inverse normal table	628	(1)
E: Critical values for X(2) statistic	629	(1)
F: Values of the Durbin-Watson statistic	630	(2)
G: Normally distributed observations	632	(1)
AUTHOR INDEX	633	(3)
SUBJECT INDEX	636