For both methods, each of the components was scaled according to the maximum and minimum 95% fractal of the global data in question. This means, that five percent of the highest and lowest values were censored. Interestingly, many of the data points fell in those extreme bounds, and in those situations, the scaled values are slightly above 1 or below 0. Censoring was done in order to avoid one or few outliers from having a dramatic on the scaling of allother data point globally.