R语言刷题检验数据缺失类型过程详解

library(tidyr)    # 绘图所需
library(ggplot2)  # 绘图所需
# 生成数据
GenerateData <- function(a = 0, b = 0, seed = 2018) {
  set.seed(seed)
  z1 <- rnorm(100)
  z2 <- rnorm(100)
  z3 <- rnorm(100)
  y1 <- 1 + z1
  y2 <- 5 + 2 * z1 + z2
  u <- a * (y1 - 1) + b * (y2 - 5) + z3
  m2 <- 1 * (u < 0)
  y2_na <- y2
  y2_na[u < 0] <- NA
  # y2_na[as.logical(m2)] <- NA
  dat_comp <- data.frame(y1 = y1, y2 = y2)
  dat_incomp <- data.frame(y1 = y1, y2 = y2_na)
  dat_incomp <- na.omit(dat_incomp)
  return(list(dat_comp = dat_comp, dat_incomp = dat_incomp))
}
# 展现缺失出具与未缺失数据的分布情况
PlotTwoDistribution <- function(dat) {
  p1 <- dat_comp %>%
    gather(y1, y2, key = "var", value = "value") %>%
    ggplot(aes(x = value)) +
    geom_histogram(aes(fill = factor(var), y = ..density..),
                   alpha = 0.3, colour = 'black') +
    stat_density(geom = 'line', position = 'identity', size = 1.5,
                 aes(colour = factor(var))) +
    facet_wrap(~ var, ncol = 2) +
    labs(y = '直方图与密度曲线', x = '值',
         title = '完整无缺失数据', fill = '变量') +
    theme(plot.title = element_text(hjust = 0.5)) +
    guides(color = FALSE)
  p2 <- dat_incomp %>%
    gather(y1, y2, key = "var", value = "value") %>%
    ggplot(aes(x = value)) +
    geom_histogram(aes(fill = factor(var), y = ..density..),
                   alpha = 0.3, colour = 'black') +
    stat_density(geom = 'line', position = 'identity', size = 1.5,
                 aes(colour = factor(var))) +
    facet_wrap(~ var, ncol = 2) +
    labs(y = '直方图与密度曲线', x = '值',
         title = '有缺失数据', fill = '变量') +
    theme(plot.title = element_text(hjust = 0.5)) +
    guides(color = FALSE)
  return(list(p_comp = p1, p_incomp = p2))
}

# 生成数据并查看数据样式
dat <- GenerateData(a = 0, b = 0)
dat_comp <- dat$dat_comp
dat_incomp <- dat$dat_incomp

head(dat_comp)
head(dat_incomp)

# 绘图展示
p <- PlotTwoDistribution(dat)
p$p_comp
p$p_incomp

t.test(dat_comp$y1, dat_incomp$y1)

# 生成数据并查看数据样式
dat <- GenerateData(a = 2, b = 0)
dat_comp <- dat$dat_comp
dat_incomp <- dat$dat_incomp

head(dat_comp)
head(dat_incomp)

# 绘图展示
p <- PlotTwoDistribution(dat)
p$p_comp
p$p_incomp

t.test(dat_comp$y1, dat_incomp$y1)

# 生成数据并查看数据样式
dat <- GenerateData(a = 0, b = 2)
dat_comp <- dat$dat_comp
dat_incomp <- dat$dat_incomp

head(dat_comp)
head(dat_incomp)

# 绘图展示
p <- PlotTwoDistribution(dat)
p$p_comp
p$p_incomp

t.test(dat_comp$y1, dat_incomp$y1)