问题如下：
A CFO in a median-sized US bank decides to choose distributions to model loss frequency and severity. The distribution that models severity:
选项：
A.should be sensitive to the tail of the distribution.
B.requires less data than distribution that models frequency
C.requires to measure the mean and variance of the loss data.
D.can use Poisson distribution
解释：
A is correct.
考点：parametric-based LDA
解析：Loss severity的建模需要重点关注尾部的分布。B选项，Loss severity的建模与loss freqency相比，需要更多的数据。C选项，关注尾部，因此mean和variance的衡量并不重要。D选项，Poisson distribution通常用于衡量loss freqency。
如何理解sensetive to tail 啊
我理解成了高敏感要求tail数据少（即thin），所以才没有选A。
有没有可以捋顺下我思维的套路？
