| Exploratory data analysis for some real-life data |
| Written by theoretic |
|
In Figures 3.2.5 and 3.2.15 we have graphically summarized some properties of two real-life data sets. The data underlying Figure 3.2.5 correspond to Danish fire insurance claims in millions of Danish Kroner (1985 prices). The data were communicated to us by Mette Rytgaard and correspond to the period 1980-1992, inclusively. There is a total of n = 2 493 observations. The second insurance data, presented in Figure 3.2.15, correspond to a portfolio of US industrial fire data (n = 8 043) reported over a two year period. This data set is definitely considered by the portfolio manager as “dangerous”, i.e., large claim considerations do enter substantially in the final premium calculation. A first glance at the figures and Table 3.2.14 for both data sets immediately reveals heavy-tailedness and skewedness to the right. The corresponding mean excess functions are close to a straight line which fact indicates that the underlying distributions may be modeled by Pareto-like distribution functions. |