R-ggdist - 分布和不确定性可视化

DataCharm

发布于 2021-02-22 12:02:40

5680

发布于 2021-02-22 12:02:40

今天的推文给大家介绍一个我发现的比较优秀的一个可视化R包-ggdist包，这是一个非常优秀和方便的用于绘制分布(distributions)和不确定性(uncertainty) 的可视化绘图包，详细介绍大家可以去官网查阅：ggdist官网。本期推文涉及的内容主要如下：

ggdist 主要绘图函数介绍
ggdist 实例绘图

ggdist 主要绘图函数介绍

数据生成并绘制

对生成的数据集进行的最小二乘（OLS）线性回归分析

生成数据

set.seed(5)
n = 10
n_condition = 5
ABC =
  tibble(
    condition = rep(c("A","B","C","D","E"), n),
    response = rnorm(n * 5, c(0,1,2,1,-1), 0.5)
  )
  
#绘图
ABC %>%
  ggplot(aes(x = response, y = condition)) +
  geom_point(alpha = 0.5,size=5) + 
  ylab("condition")+
labs(
   title = "ggdist Exercise",
   subtitle = "base plot",
   caption = 'Visualization by DataCharm')+
  theme(
     text = element_text(family = "Times_New_Roman",face='bold')
  )

结果如下：

我们对数据进行简单的线性拟合：

m_ABC = lm(response ~ condition, data = ABC)
#使用 tidy() 进行格式化展示
tidy(m_ABC)

结果如下：

使用stat_dist_halfeye()方法绘制

m_ABC %>%
  tidy() %>%
  ggplot(aes(y = term)) +
  stat_dist_halfeye(
    aes(dist = "student_t", arg1 = df.residual(m_ABC), arg2 = estimate, arg3 = std.error)
  )+
ylab("condition")+
labs(
   title = "ggdist Exercise",
   subtitle = "half-eye plots",
   caption = 'Visualization by DataCharm')+
  theme(
     text = element_text(family = "Times_New_Roman",face='bold')
  )

结果如下：

使用stat_dist_halfeye()方法绘制

ABC %>%
  data_grid(condition) %>%
  augment(m_ABC, newdata = ., se_fit = TRUE) %>%
  ggplot(aes(y = condition)) +
  stat_dist_halfeye(
    aes(dist = "student_t", arg1 = df.residual(m_ABC), arg2 = .fitted, arg3 = .se.fit), 
    scale = .5
  ) +
  geom_point(aes(x = response), data = ABC, pch = "|", size = 2, position = position_nudge(y = -.15))+

ylab("condition")+
labs(
   title = "ggdist Exercise",
   subtitle = "dist-halfeye plots",
   caption = 'Visualization by DataCharm')+
  theme(
     text = element_text(family = "Times_New_Roman",face='bold')
  )

结果如下：

使用 stat_dist_gradientinterval()方法绘制

ABC %>%
  data_grid(condition) %>%
  augment(m_ABC, newdata = ., se_fit = TRUE) %>%
  ggplot(aes(y = condition)) +
  stat_dist_gradientinterval(
    aes(dist = "student_t", arg1 = df.residual(m_ABC), arg2 = .fitted, arg3 = .se.fit), 
    scale = .5
  )+

ylab("condition")+
labs(
   title = "ggdist Exercise",
   subtitle = "gradientinterval plots",
   caption = 'Visualization by DataCharm')+
  theme(
     text = element_text(family = "Times_New_Roman",face='bold')
  )

结果如下：

使用stat_dist_ccdfinterval()方法绘制

ABC %>%
  data_grid(condition) %>%
  augment(m_ABC, newdata = ., se_fit = TRUE) %>%
  ggplot(aes(y = condition)) +
  stat_dist_ccdfinterval(
    aes(dist = "student_t", arg1 = df.residual(m_ABC), arg2 = .fitted, arg3 = .se.fit)
  )+

ylab("condition")+
labs(
   title = "ggdist Exercise",
   subtitle = "CCDF plots",
   caption = 'Visualization by DataCharm')+
  theme(
     text = element_text(family = "Times_New_Roman",face='bold')
  )

结果如下：

以上就是结合ggdist官网的例子对其基本绘图函数进行简单的可视化效果展示，更多其他绘制方法，大家可以阅读官网的例子。

ggdist 实例绘图

实例展示部分涉及一些基本的 ggplot2的绘图函数以及基本的数据操作部分，这里我们直接给出代码(部分重点部分会给出解释)

绘图原始数据格式如下(部分)

数据处理

df_rect <-
  tibble(
    xmin = c(-Inf, 2.46, 3.27),
    xmax = c(Inf, Inf, Inf),
    ymin = c(3, 2, 1),
    ymax = c(Inf, Inf, Inf)
  )

df_peng_iqr <- 
  df_penguins %>% 
  mutate(bill_ratio = bill_length_mm / bill_depth_mm) %>% 
  filter(!is.na(bill_ratio)) %>% 
  group_by(species) %>% 
  mutate(
    median = median(bill_ratio),
    q25 = quantile(bill_ratio, probs = .25),
    q75 = quantile(bill_ratio, probs = .75),
    n = n()
  ) %>% 
  ungroup() %>% 
  mutate(species_num = as.numeric(fct_rev(species)))

数据可视化绘制

这里直接给出可视化绘制代码(都是基本的绘图函数)

test_plot <- 
  ggplot(df_peng_iqr, aes(bill_ratio, species_num - .2)) +

geom_linerange(
    data = df_peng_iqr %>% 
      group_by(species, species_num) %>% 
      summarize(m = unique(median)),
    aes(
      xmin = -Inf, 
      xmax = m, 
      y = species_num,
      color = species
    ),
    inherit.aes = F,
    linetype = "dotted",
    size = .7
  ) +
  geom_boxplot(
    aes(
      color = species,
      color = after_scale(darken(color, .1, space = "HLS"))
    ),
    width = 0,
    size = .9
  ) +
  geom_rect(
    aes(
      xmin = q25,
      xmax = median,
      ymin = species_num - .05,
      ymax = species_num - .35
    ),
    fill = "grey89"
  ) +
  geom_rect(aes(xmin = q75,xmax = median,ymin = species_num - .05,ymax = species_num - .35),
    fill = "grey79"
  ) +
  geom_segment(
    aes(
      x = q25, 
      xend = q25,
      y = species_num - .05,
      yend = species_num - .35,
      color = species,
      color = after_scale(darken(color, .05, space = "HLS"))
    ),
    size = .25
  ) +
  geom_segment(aes(
      x = q75, 
      xend = q75,
      y = species_num - .05,
      yend = species_num - .35,
      color = species,
      color = after_scale(darken(color, .05, space = "HLS"))
    ),
    size = .25
  ) +
geom_point(aes(color = species), shape = "|",size = 5,alpha = .4) +
  ggdist::stat_halfeye(aes(y = species_num,color = species,fill = after_scale(lighten(color, .5))),
    shape = 18,
    point_size = 3,
    interval_size = 1.8,
    adjust = .5,
    .width = c(0, 1)
  ) +
scale_x_continuous(
    limits = c(1.57, 3.7),
    breaks = seq(1.6, 3.6, by = .2),
    expand = c(.001, .001)
  ) +
  #对y轴刻度进行定制化涉及
  scale_y_continuous(
    limits = c(.55, NA),
    breaks = 1:3,
    labels = c("Gentoo", "Chinstrap", "Adélie"),
    expand = c(0, 0)
  ) +
  #映射颜色设置
  scale_color_manual(
    values = pal,
    guide = F
  ) +
 #映射填充(fill)设置
  scale_fill_manual(
    values = pal,
    guide = F
  )  +
labs(
    x = "Bill ratio",
    y = NULL,
    title = "ggdist Exercise",
    caption = 'Visualization by DataCharm'
  ) +
 #绘图主题定制化设置
theme(
  panel.grid.major = element_line(color = "grey92", size = .4),
  panel.grid.major.y = element_blank(),
  panel.grid.minor = element_blank(),
  axis.title.x = element_text(color = "grey30", margin = margin(t = 7)),
  axis.title.y = element_text(color = "grey30", margin = margin(r = 7)),
  axis.text = element_text(color = "grey50"),
  axis.text.y = element_text(family = "IBMPSBold", 
                               color = rev(pal), size = 14, lineheight = .9),
  axis.ticks =  element_line(color = "grey92", size = .4),
  axis.ticks.length = unit(0, "lines"),
  plot.title = element_text(hjust = 0, color = "black", 
                            family = "IBMPSBold",
                            size = 21, margin = margin(t = 10, b = 35)),
  plot.caption = element_text(color = "grey50", size = 10, hjust = 1,
                              family = "IBMPSBold", 
                              lineheight = 1.05, margin = margin(30, 0, 0, 0)),
  plot.margin = margin(rep(20, 4))
)
test_plot

可视化结果如下：