We will use simulation to evaluate how well can a normal distribution approximate a binomial distribution. Suppose that X ~ Binomial(n,p) . A theorem says that if is large, then T = 根号n * (X/n-p)is approximately N(0,p(1-p)). Generate a sample of 100 binomial distributions each time. Use the Kolmogorov-Smirnov test to evaluate whether the deviation of T from N(0,P(1-P)) can be detected at the alpha = 0.05level(you may use ks.test). experiment with n belong {10,50,100,200} and p belong {0.01,0.1, 0.5}
没有任何思路，理解不来逻辑，更不知道怎么写