R | 统计连续变量在各个区间的频次

我们通常会通过直方图来观察数据的结构，但对于如何在R中实际地统计各个区间的数据频次却较少接触。因此，本文总结三种将连续变量离散化的操作：

ggplot2::cut_width()
ggplot2::cut_interval()
ggplot2::cut_number()

数据模拟

首先，我们从正态分布模拟100个数据进行后续分析

set.seed(123)
dat1 <- tibble(Num=rnorm(100))

cut_width

cut_width()将数据切成长度为width的区间

dat1 %>% 
  mutate(Interval=cut_width(dat1$Num, width=0.5)) %>% 
  count(Interval)

# A tibble: 10 x 2
   Interval          n
   <fct>         <int>
 1 [-2.75,-2.25]     1
 2 (-2.25,-1.75]     1
 3 (-1.75,-1.25]     4
 4 (-1.25,-0.75]     8
 5 (-0.75,-0.25]    23
 6 (-0.25,0.25]     21
 7 (0.25,0.75]      18
 8 (0.75,1.25]      13
 9 (1.25,1.75]       7
10 (1.75,2.25]       4

dat1 %>% 
  mutate(Interval=cut_width(dat1$Num, width=0.5)) %>% 
  count(Interval) %>% 
  ggplot(aes(x=Interval, y=n)) + 
  geom_bar(stat='identity')

cut_interval

cut_interval()将数据切成 n 个区间

dat1 %>% 
  mutate(Interval=cut_interval(dat1$Num, n=10)) %>% 
  count(Interval)

# A tibble: 10 x 2
   Interval             n
   <fct>            <int>
 1 [-2.31,-1.86]        2
 2 (-1.86,-1.41]        2
 3 (-1.41,-0.96]       10
 4 (-0.96,-0.511]      10
 5 (-0.511,-0.0609]    22
 6 (-0.0609,0.389]     18
 7 (0.389,0.838]       15
 8 (0.838,1.29]        11
 9 (1.29,1.74]          6
10 (1.74,2.19]          4

cut_number

cut_number()将数据等分，确保每个区间有n个数据。类似于划分为n分位数。

dat1 %>% 
  mutate(Interval=cut_number(dat1$Num, n=10)) %>% 
  count(Interval)

# A tibble: 10 x 2
   Interval            n
   <fct>           <int>
 1 [-2.31,-1.07]      10
 2 (-1.07,-0.626]     10
 3 (-0.626,-0.387]    10
 4 (-0.387,-0.223]    10
 5 (-0.223,0.0618]    10
 6 (0.0618,0.315]     10
 7 (0.315,0.513]      10
 8 (0.513,0.882]      10
 9 (0.882,1.26]       10
10 (1.26,2.19]        10

Ref:

https://ggplot2.tidyverse.org/reference/cut_interval.html

完。