The final exam results of students A and B are A: Math 90, Chinese 90 and common sense 60; B: Math 80, Chinese 80 and common sense 80. Now ask you AB who studies well? How do you judge? If we simply calculate the average of the two, we find that they are the same, but in fact, their scores are still biased. However, the numerical characteristics of the average can no longer be judged. At this time, we assume that child AB has 20 classes every week, including 8 classes in mathematics, 2 classes in Chinese 10 and 2 classes in common sense, so that we let E (a) = 90 * 8/20+90 *10/20+60 * 2/20; E (b) = 80 * 8/20+80 *120+80 *10/20, E (a) >: E(B) shows that the quality of AB can be judged under this algorithm, that is, e (x) = ∑ xi * pi. In short, expectation is a generalization of an average, and the average is of equal probability, so different weights are considered for expectation. Of course, it was later found that random variables are still biased under the condition of equal expectations, which leads to other numerical characteristics such as variance.