Preface

Most statistical applications involve computational work with data stored on a computer. The mechanics of interaction with the data is a function of the statistical computing environment. This application guide is intended for slightly experienced statisticians in computer-aided data analysis who desire to learn advanced applications in various fields of statistics. The prerequisities for XploRe --the statistic computing environment-­are an introductory course in statistics or mathematics. This book is designed as an e-book which means that the text contained in here is also available as an integrated document in HTML and PDF format. The reader of this application guide should therefore be familiar with the basics of Acrobat Reader and of HTML browsers in order to profit from direct computing possibilities within this document.

The quantlets presented here may be used together with the academic edition of XploRe ( http://www.i-xplore.de ) or via the XploRe Quantlet Client (XQC) on http://www.xplore-stat.de . The book comes together with a CD-Rom that contains the XploRe Quantlet Server (XQS) and the full Auto Pilot Support System (APSS). With this e-book bundle one may directly try the application without being dependent on a specific software version.

The quantlets described in the book can be accessed via the links included in the text. All executable quantlets are denoted by the symbol 1692 XAGclust01.xpl . Some quantlets need more time to finish and we denote them by the symbol 1695 XAGclust01.xpl . Quantlets which require the professional edition of XploRe (more than 1000 observations) are denoted by 1702 XAGclust01.xpl .

The XploRe language is intuitive and users with prior experience of other statistical programs will find it easy to reproduce the examples explained in this XploRe--Application Guide (XAG). The XAG may be seen as a complement to the Springer XploRe--Learning Guide (XLG) but can be used independently. The quantlets of this e-book are also available on the internet. The XploRe language is described in detail in the XploRe--Reference Guide (XRG), http://www.xplore-stat.de . The statistical applications that the reader is guided through range from a discussion on regression problems to more complicated tasks such as additive modelling, hazard regression, classification and regression trees, Kalman filtering, partial least squares etc.

Accordingly, the XAG is divided into three main parts: Regression Models, Data Exploration, and Dynamic Statistical Systems.

In the first part the XAG starts with quantile regression and the application to robustified regression estimation. The least trimmed squared technique is then applied to a data set of phone calls. The regression with errors in the variables is shown for the analysis of agricultural data. The simultaneous equation model and its estimation by two-stage least squares with application to money demand is presented before the chapter on hazard repression. The quantlib hazreg provides methods for analysing right-censored time-to-event data. This technique is applied to the analysis of length of stay in nursing homes. The gplm quantlib introduces into generalised partially linear modelling. An application to credit scoring is presented here. The first part ends with a chapter on generalised additive modelling from the quantlib gam .

The second part is devoted to data exploration and starts with an analysis on income dynamics and poverty traps based on income distribution data from 1960-1985. Applications of cluster analysis to discriminating between forged and genuine banknotes are presented next. The CART technology is used in the analysis of the Boston housing data set. Dynamic partial least squares is used in considering the dynamics of German share prices. The second part closes with a panel data study on uncovered interest parity and a tool for contingency tables based on correspondence analysis.

The last part starts with the presentation of long memory quantlets and flexible time series analysis. The multiple time series quantlets are shown in action for a money demand system. The XAG ends with a chapter on robust Kalman filtering.

XploRe and this XAG have benefited at several stages from co-operation with colleagues and students. We want to mention in particular Marlene Müller, Torsten Kleinow, Heiko Lehmann, Bernd Rönz, Michal Benko, Sven Denkert, Jörg Feuerhake, Petr Franek, Christian Hafner, Christian Hipp, Joel Horowitz, Roger Koenker, Thomas Kühn, Danilo Mercurio, Fabian Nötzel, Erich Neuwirth, Dirk Schnell, Léopold Simar, Rodrigo Witzel, Uwe Ziegenhagen.

We owe special thanks to Clemens Heine, Springer Verlag for professional editorship and good directions at critical junctions of this e-book.

W. Härdle, Z. Hlávka, S. Klinke

Berlin, May 2000

XploRe, MD*Tech and Quantlets are registered trademarks