Combining different methodologies in the field of soft computing as rule based fuzzy classification systems, fuzzy clustering methods and neural networks like multilayer perceptrons, enables us to combine their advantages. As this work shows that the well known and commonly used fuzzy max-min classification system offers only restricted usability, mainly Lukasiewicz classification systems are considered. Those systems are more flexible being capable of solving piecewise linear problems. The classification methodology of such a fuzzy classification system is geometrically characterized and visualized. The same is presented for fuzzy clustering systems and multilayer perceptrons. On one hand this visualization enables the user to initialize a multilayer perceptron with the information gained from a fuzzy clustering system or a fuzzy rule based system, on the other hand fuzzy rules can be deducted from a fuzzy clustering system without loosing information. The developed methodology is applied in the field of air traffic control. It is used to improve prediction of delay times on airports.