非参数假设检验——Pearson检验法

最新推荐文章于 2025-02-23 19:25:42 发布

LS_FIGHTING

最新推荐文章于 2025-02-23 19:25:42 发布

阅读量8.5k

点赞数 3

本文链接：https://blog.csdn.net/LEE_FIGHTING_JINGYU/article/details/105837548

版权

1. 非参数假设检验

常常事先并不知道总体的分布类型，这就要根据抽取样本所提供的信息，对总体分布的各种假设进行检验。称总体分布未知时所进行的假设检验为非参数假设检验。

2. Pearson检验法 / $\chi ^2$ 拟合检验法

3. 在R语言中可以利用chisq.test()函数来实现

chisq.test(x, y = NULL, correct = TRUE,
           p = rep(1/length(x), length(x)), rescale.p = FALSE,
           simulate.p.value = FALSE, B = 2000)

4.例题1

zhengchang<-c(535,382)
semang<-c(65,18)
chisq.test(rbind(zhengchang,semang),correct = FALSE)

运行界面：

例题2

data<-c(459,362,624,542,509,584,433,748,815,505,612,452,434,982,640,742,565,706,593,680,926,653,164,487,734,608,428,1153,593,844,527,552,
     513,781,474,388,824,538,862,659,775,859,755,649,697,515,628,954,771,609,402,960,885,610,292,837,473,677,358,638,699,634,555,
     570,84,416,606,1062,484,120,447,654,564,339,280,246,687,539,790,581,621,724,531,512,557,496,468,499,544,645,764,558,378,765,666,763,
     217,715,310,851)
# 输出向量（最小值，最大值）
range(data)
m<-84
n<-1153
# seq(from,to,length.out=by)表示生成一组从from到to的数量为num的数
# by = ((to - from)/(length.out - 1))
breaks<-seq(m,n,length.out = 9)
xx<-cut(data,breaks = breaks)
freq <-table(xx)
p <- c(0, pnorm(breaks[-c(1, length(breaks))], mean = mean(data), sd = sd(data)), 1)
p <- p[-1] - p[-length(p)]
e <- p*100
chi_square <- sum((freq - e)^2/e)
p_value <- 1 - pchisq(chi_square, length(breaks) - 1 - 3)