Statistically matching of separate survey samples - can this be efficient? When there is no single source file available about all the information of interest, techniques of matching different data sets are often applied. Then individual respondents on one survey are matched to those on another based on some common characteristics. The respondents in the resulting data set will have all the answers to all the questions in both original surveys. For example, government policy questions as well as media planning tasks may be answered by means of such a statistically matched data set. This book covers a wide range of different aspects concerning statistical matching that in Europe typically is called data fusion. A theoretical framework is derived to determine the advantages and disadvantages of statistical matching. Its history and practical applications are discussed, and alternative approaches are proposed and evaluated with real world marketing data. Answers to the question of efficiency are provided. A book about statistical matching will be of interest to researchers and practitioners in many data analysis areas, starting with data collection and the production of public use micro files, data banks, and data bases. People in the areas of database marketing, public health analysis, socioeconomic modeling, and also official statistics also will find it useful. Susanne Rässler is senior research assistant and lecturer at the Institute of Statistics and Econometrics at the University of Erlangen-Nürnberg in Germany. She received her Ph.D. in 1995, having written a book about survey sampling theory with the focus on sampling with unequal probabilities. Later she started research about statistical matching. This book is the result of her "habilitation" thesis according to German academic tradition.

Mit diesem Buch liegen kompakte Beschreibungen von Prognoseverfahren vor, die vor allem in Systemen der betrieblichen Informationsverarbeitung eingesetzt werden. Praktiker mit langjahriger Prognoseerfahrung zeigen ausserdem, wie die einzelnen Methoden in der Unternehmung Verwendung finden und wo die Probleme beim Einsatz liegen. Das Buch wendet sich gleichermassen an Wissenschaft und Praxis. Das Spektrum reicht von einfachen Verfahren der Vorhersage uber neuere Ansatze der kunstlichen Intelligenz und Zeitreihenanalyse bis hin zur Prognose von Softwarezuverlassigkeit und zur kooperativen Vorhersage in Liefernetzen. In der siebenten, wesentlich uberarbeiteten und erweiterten Auflage werden neue Vergleiche von Prognosemethoden, GARCH-Modelle zur Finanzmarktprognose, Predictive Analytics" als Variante der Business Intelligence" und die Kombination von Vorhersagen mit Elementen der Chaostheorie berucksichtigt."