Astron. Astrophys. 323, 513-523 (1997)

Appendix A: list of class I objects

Appendix B: list of class II objects

Appendix C: the boxplot

A boxplot provides information on the location, spread, skewness, tail length and outliers of a data set. It is a powerful exploratory tool, especially when several data sets are to be compared simultaneously. Boxplots replace the more often used histograms, which are very much biased by the choice of the bin size (Tukey 1990). As can be seen from Fig.3, a boxplot depicts 5 quantities: the minimum and maximum value of the data set, labeled min and max, the first and third quartile, labeled Q1 and Q3 and the median. Their meaning is the following. When the data are put in ascending order, the first 25% are located between min and Q1, the next 25% between Q1 and the median, etc. So, 75% of the data are smaller than the value of Q3.

An indication for the spread of the data is given by the Inter Quartile Range, IQR, defined as . It is the length of the box, containing 50% of the data. In case of a data set containing outliers, i.e. data points with extreme high or low values in comparison with the bulk of the data, the IQR provides a better indication of the spread than the standard deviation, since the latter can be strongly influenced by outliers.

In Table 8, we list the values for these boxplot parameters for both the excess and the Rayleigh-Jeans sample.

More information on boxplots can be found in Tukey (1977).

Table 8. Boxplot parameters for the two samples

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998

helpdesk.link@springer.de