This book describes the fundamental methodological aspects of the analysis and modelling of spacially distributed data, and the applications with the specific userfriendly software Geostat Office. The methods presented in this book include two domains of geostatistics and of machine learning algorithms, and some aspects of Geographical Information Systems. The geostatistical methods cover the traditionnal variography and spatial predictions, as well as an extensive part on conditional stochastic simulations and estimation of local probability distribution functions. A special chapter is devoted to the exploratory spatial data analysis, where the analysis of monitoring network is extensively decribed. In addition to more traditional geostatistics, the methods of artificial neural networks of different architectures ans Support Vector Machines (SVM) are explained ans illustrated. The key feature of machine learning algorithms is that learn from data and can be efficiently used when the modelled phenomenon is not described accurately. Machine Learning algorithms are adaptive tools to solve prediction, characterization, optimisation and density estimation problems. The fundamentals of Statistical Learning Theory (Vapnik-Chervonenkis theory) is explained using examples of real environmental spatial data; SVM develop robust data models with good generalisation capabilities. The book is distributed with the student version of Geostat Office Software which runs under Microsoft Windows. The book and its GSO software can be useful for teaching as well as for modelling real case studies.::This book may be ordered by customers located in Switzerland, France,Belgium and North Africa only. For other countries, please contactDekker Ltd. at www.dekker.com::