|Visual data mining multi-dimensional scaling POLARMAP Sammon's mapping clustering outlier detection|
Almost all branches of commerce, industry and research put great efforts in collecting data with the objective to describe and predict customer behaviour or both technical and natural phenomena. Besides the size of such data sets, which make manual analysis impractical, data analysis becomes challenging due to a large number of attributes describing a data object. Whereas a graphical representation of data objects that are described by means of two or three attributes can be realized easily, the visualization of high-dimensional data -- data that is described through many attributes -- is not trivial. The data mining research area comprises the development of suitable techniques for data preprocessing and data analysis to cope with the problem of aggrandizing databases including complex data sets. This thesis contributes to the domain of methodology development, dimensionality reduction and outlier treatment. Another major focus is set on the visualization of complex data as well as the visualization of complex results obtained from common data mining techniques, e.g. clustering and fuzzy classifiers. The characteristics of the proposed techniques become evident on the example of the analysis of flight data and weather data measured at Frankfurt Airport. The objective of this application is the research of weather factors that affect the flight duration of aircraft approaching Frankfurt Airport. Understanding the interrelationship between weather and flight duration permits the optimization of various processes at the respective airport and may save time and money of customers and companies.