Data mining approach for environmental data predictions and classification

摘要

The current work presents application of several data mining approaches - different kinds of Artificial Neural Networks, Support Vector Machines (SVM) and geostastics - for classification and regression case studies (soil types classification, hydrogeological parameter zonation, spatio-temporal prediction of ground-water contamination). The main purpose of the current work is to attract attention to data mining capabilities concerning different aspects of data analysis: uncertainty assessment, incorporation of additional information, feature extraction, validation of methods, etc. Each case study is organized so as to demonstrate some special features of the applied methods. The main conclusion is that application of data mining methods can be useful for different research areas when dealing with huge amounts of data presenting complex spatial/temporal features.