For example, let us consider a special case where only one attribute, income, is
used for classification, and suppose we need to classify a person whose income is
76000.We assume that income values are broken up into buckets, and assume that
the bucket containing 76000 contains values in the range (75000, 80000). Suppose
among instances of class excellent, the probability of income being in (75000,
80000) is 0.1, while among instances of class good, the probability of income being
in (75000, 80000) is 0.05. Suppose also that overall 0.1 fraction of people are
classified as excellent, and 0.3 are classified as good. Then, p(d|c j )p(c j) for class
excellent is .01, while for class good, it is 0.015. The person would therefore be
classified in class good.