Mildenhall Aggregate
Loss Tools
Home
Contents
Back
Next
Help
Example 4: Computing Frequency Distributions
Distribution Specification
Frequency Distribution
Poisson
Negative Binomial
Poisson-Inverse Gaussian
Num L_2
Mean
Num L_2 = log_2 length of output vector
VarMult
Click compute to see results...
Notes
If severity X is the degenerate distribution Pr(X=1) = 1, the the aggregate distribution is identicial to the underlying frequency distribution.
This example uses a degenerate severity distribution to compute frequency distributions.
Percentage errors from the exact solution are also shown.
Errors in the FFT method are caused by aliasing. They can be reduced by selecting a larger number of buckets (larger value for num_l2).
References to Wang
Section 3.3.1, page 865.
See Section 2.3.2 for a discussion of aliasing, where "tail probabilities wrap around and appear at the beginning."