r语言-连续型变量组间差异的比较

## 连续型变量组间差异的比较 ##

getwd()

setwd("C:/我的/广西医科大学/专业核心课/R语言课程/6.17")

##1.单样本t检验##

#某山区汉族足月正常产男婴的双顶径是否大于一般男婴的双顶径(9.3cm)

mydata<-c(9.95,9.33,9.49,9.00,9.90,9.15,9.52,9.33,9.16,9.37,9.11,9.27) # 构建数据集

shapiro.test(mydata) #小样本正态性判定

qqnorm(mydata) # 绘制QQ图

qqline(mydata, col=2, lwd=1)

t.test(mydata, alternative = "greater", mu = 9.3) # 设置参数alt进行单侧检验

##2.独立样本t检验##

#比较两组小鼠肉芽肿的重量是否有差异

A1<-c(108,74.8,31.2,132 ,147.6,98.5,82.2,93.3,85.5,110.4)

A2<-c(94.8,122.5,144.1,151.2,189.3,204.2,155.6,160.3,178.3,165.4)

t.test(A1,A2, var.equal=TRUE)

#比较吸烟组和不吸烟组产妇的新生儿体重是否有差异

data(birthwt, package="MASS")

var.test(bwt ~ smoke, data=birthwt) # 方差齐性检验，p=0.2254>0.05，差异无统计学意义，方差齐

t.test(bwt ~ smoke, var.equal=TRUE, data=birthwt) # 参数设置=TRUE表明方差齐，默认是FALSE

                                                  # p<0.05，新生儿体重在吸烟的母亲和不吸烟的母亲之间差异有统计学意义

t.test(bwt ~ smoke, var.equal=TRUE, conf.level=0.99, data=birthwt) # 调整参数conf.level设置置信度

##3.配对样本t检验##

x1 <- c(0.84, 0.59, 0.67, 0.63, 0.69, 0.98, 0.75, 0.73, 1.20, 0.87) # 脂肪酸水解法测定10份乳酸菌饮料中脂肪含量测定的结果

x2 <- c(0.58, 0.51, 0.50, 0.32, 0.34, 0.52, 0.45, 0.51, 1.00, 0.51) # 罗紫-戈特里法测定结果

t.test(x1,x2, paired = TRUE) # 配对T检验参数设置为TRUE

                           # p<0.001, 两种方法对脂肪含量的测定的差异有统计学意义

##4.单因素方差分析（多组比较）##

#不同组之间植物重量是否存在显著性差异

install.packages("PlantGrowth")

summary(PlantGrowth)

boxplot(weight ~ group, data = PlantGrowth)# 制作箱形图

tapply(PlantGrowth$weight,PlantGrowth$group,shapiro.test)#正态性检验. 用函数tapply进行分组计算

bartlett.test(weight ~ group, data = PlantGrowth)#方差齐性检验(正态性敏感)

## Levene检验

install.packages("car")

library(car)

leveneTest(weight ~ group, data = PlantGrowth) # 方差齐性检验

mymodel <- aov(weight ~ group, data = PlantGrowth) #建立方差分析模型

summary(mymodel) # 得到方差分析表

TukeyHSD(mymodel)# 进行多重比较

plot(TukeyHSD(mymodel))

##5.组间差异的非参数检验##

wilcox.test(A1,A2)#两独立样本

wilcox.test(bwt ~ smoke, data = birthwt)

wilcox.test(x1,x2, paired = TRUE) # 配对样本

kruskal.test(weight ~ group, data = PlantGrowth)

install.packages("PMCMRplus")

library("PMCMRplus")

comp<-bwsAllPairsTest(weight ~ group, data = PlantGrowth)#多重比较

summary(comp)

##6.函数tablestack() 汇总双变量分析结果##

install.packages("MASS")

library("MASS")

library(dplyr)

#变量转化为因子型

str(birthwt)

birthwt <- birthwt %>%

  mutate(low=factor(low,labels = c("no","yes")),

         race=factor(race,labels=c("white","black","others")),

         smoke=factor(smoke,labels = c("no","yes")),

         ht=factor(ht,labels = c("no","yes")),

         ui=factor(ui,labels = c("no","yes")))

str(birthwt)

install.packages("epiDisplay")

library("epiDisplay")

#如果我们关注的是新生儿低体重，那么因子型变量low就是结果变量

tableStack(vars=age:smoke,by=low,dataFrame=birthwt) # 创建包含多个变量的统计汇总表

tableStack(vars = c(age,smoke),by=low,dataFrame=birthwt) # 自行选择需要的变量进行统计汇总

Data<-tableStack(vars=age:smoke,by=low,dataFrame=birthwt)

write.csv(Data,file= "C:/我的/广西医科大学/专业核心课/R语言课程/Data.csv") #保存