In cases like this, where the independent variable does not have an intrinsic ordering, it is
often a good idea to sort entries by the dependent variable. That’s what we did in the
example: rather than defining some (arbitrary) sort order on the job titles, we sorted by
the number of records (i.e., by the dependent variable). Once the records have been sorted
in this way, we can form a histogram and a CDF as before
实质是变量没有内在联系，无法排序。所以用变量的数值进行排序。同时生成一个直方图和累计分布图。
PIC2, Rank-Order Plots, Lift Charts and Pareto chart
原文：http://www.cnblogs.com/hluo/p/4047810.html