Treatment of Measurement Results

The processing of measurement results from technical data analysis is usually based on the a priori assumptions about their statistical behavior as independence, absence of outliers, and normality. These assumptions are used in all classical statistical methods. For smaller samples and imprecise measurements, simple tests of assumptions about data and robust techniques are used. This chapter is devoted to combining classical methods with data-oriented exploratory methods capable of characterizing individual data behavior by using proper graphical tools. The data transformation and computer-intensive methods for data treatment in cases of violation of basic assumptions and for data far from a normal distribution are described in detail.