Applied Data Mining for Forecasting Reviews

"This well-organized and well-written book is unusual in that it takes you through the complete forecasting process—from the beginning planning stages through data collection, cleaning, and final analysis with a nice summary example in the last chapter that ties everything together. Discussion of the many decisions you'll need to make that were not covered in your statistics textbooks make this an especially useful reference for those who actually do forecasting. It covers a lot of ground from simple to quite sophisticated modeling techniques without excessive mathematical detail."

David Dickey
William Neal Reynolds Distinguished Professor of Statistics
North Carolina State University

"I believe this is an excellent book for frontier practitioners and researchers, especially in the forecasting and data mining fields. Leveraging data mining techniques in an era where economic data and industry trend data are ample and readily available for building casual-effect forecasting models is a very promising endeavor. With a good causal-effect predictive model built through this approach, businesses would know what external and internal factors are truly driving effects and scenario-based forecasts and hence various contingency plans could be made in a timely fashion. I applaud the great effort by the authors."

Jerry Z. Shan
Principal Scientist, HP Labs
Hewlett-Packard Company

"This book is a must for practitioners who use SAS and who deal with forecasting issues based on time-series data. The underlying purpose of this book is to provide applications for the use of data mining for forecasting purposes so as to leverage the numerous sources of time-series data readily available to decision makers. In turn, this leveraging provides a competitive advantage to decision-makers via improvements in forecast performance. The text consists of twelve chapters related to the following topics: why industries needs data mining for forecasting; the process, infrastructure, and application issues associated with data mining for forecasting; data collection and preparation; methods for variable reduction and variable selection; the construction of autoregressive and moving-average (ARMA) models; the construction of unobservable component models (UCM), and the construction of dynamic regression (ARIMAX) models. Moreover, the text provides discussions of forecasting models based on neural networks, support vector machines, evolutionary computation, and vector autoregression. Additional topics include creating data hierarchies, statistical forecast reconciliation, intermittent demand, high-frequency data and mixed-frequency forecasting, the use of holdout samples and forecast model selection, scenario-based forecasting, and new product forecasting. The key component in each of the respective chapters is the emphasis on applications with the use of examples. In fact, the final chapter of the text pulls together all the pieces of the process in systematic fashion through the provision of a detailed example of data mining for forecasting.

The book lays out the work process for the development, the deployment, and the maintenance of a forecasting project. In this process, issues indigenous to data collection, data preparation steps (consolidation of data from different sources, noisy data, data glitches, missing data, interpolations of data, data outliers, and data transformations), and data mining methods (stepwise regression, principal component analysis, clustering algorithms, decision trees and genetic programming) are described in detail. Most textbooks in this subject matter area do not spend enough time on these basic but important topics. Subsequently, attention is centered on the strengths and limitations of univariate and multivariate forecasting methods. Another differentiated product provided by this book is the ability to understand the use of various forecasting methods without getting lost in the mathematical details. As well, in this treatment, a plethora of easy-to-follow examples done in SAS are provided. SAS software tools related to data mining in forecasting (SAS Enterprise Guide, SAS Enterprise Miner, SAS Forecast Server, SAS Forecast Studio, and SAS/ETS software) are discussed at length as well.

Simply put, Applied Data Mining for Forecasting Using SAS, written by Rey, Kordon, and Wells, adds much to the literature on the topic of forecasting. Its applied, data-driven focus makes this book amenable to practitioners. Additionally, this book should be adopted in the academic community, especially in graduate courses that focus on forecasting."

Oral Capps, Jr.
Regents Professor, Executive Professor, and Co-Director
The Agribusiness, Food and Consumer Economics Research Center (AFCERC)
Texas A&M University

Back to Top