重塑一个数据集

楚新元 / 2025-07-09

本文是《R 语言实战》（第 2 版）中数据整合与重构的一个例子，书中用的是老旧的 reshape2 对数据进行了融合和重铸，有点像《变脸》里唱的那样，文化领域百花齐放，百家争鸣是对的，可是一张表表达的意思是一个意思，不同的人做出来就可能千差万别，如果你不能做到源头控制，那你就得掌握如何通过各种变换变成你需要的那张脸。

加载相关 R 包

library(tidyr)
library(dplyr)
library(knitr)

原始数据集

mydata = data.frame(
  ID = c(1, 1, 2, 2),
  Time = c(1, 2, 1, 2),
  X1 = c(5, 3, 6, 2),
  X2 = c(6, 5, 1, 4)
)
kable(mydata, align = "c")

ID	Time	X1	X2
1	1	5	6
1	2	3	5
2	1	6	1
2	2	2	4

原表转化为长表

mydata |> 
  pivot_longer(
    X1:X2,
    names_to = "variable",
    values_to = "value"
  ) -> md
kable(md, align = "c")

ID	Time	variable	value
1	1	X1	5
1	1	X2	6
1	2	X1	3
1	2	X2	5
2	1	X1	6
2	1	X2	1
2	2	X1	2
2	2	X2	4

执行整合

转化会导致信息损失，就像麦子变成面粉一样，这就意味着转换后的表无法逆回到之前的状态。

长表转化为表 a

md |> 
  summarise(
    value_mean = mean(value),
    .by = c(ID, variable)
  ) |> 
  pivot_wider(
    names_from = variable,
    values_from = value_mean
  ) -> table_a
kable(table_a, align = "c")

ID	X1	X2
1	4	5.5
2	4	2.5

长表转化为表 b

md |> 
  summarise(
    value_mean = mean(value),
    .by = c(Time, variable)
  ) |> 
  pivot_wider(
    names_from = variable,
    values_from = value_mean
  ) -> table_b
kable(table_b, align = "c")

Time	X1	X2
1	5.5	3.5
2	2.5	4.5

长表转化为表 c

md |> 
  summarise(
    value_mean = mean(value),
    .by = c(ID, Time)
  ) |> 
  pivot_wider(
    names_from = Time,
    names_prefix = "Time",
    values_from = value_mean
  ) -> table_c
kable(table_c, align = "c")

ID	Time1	Time2
1	5.5	4
2	3.5	3

不执行整合

这就意味着信息必须是无损的，只是做了形式转换，转换后可以逆回到原来的状态。

长表转为表 d

md |> 
  pivot_wider(
    names_from = variable,
    values_from = value
  ) -> table_d
kable(table_d, align = "c")

ID	Time	X1	X2
1	1	5	6
1	2	3	5
2	1	6	1
2	2	2	4

表 d 逆回长表

table_d |> 
  pivot_longer(
    X1:X2,
    names_to = "variable",
    values_to = "value"
  ) |> 
  kable(align = "c")

ID	Time	variable	value
1	1	X1	5
1	1	X2	6
1	2	X1	3
1	2	X2	5
2	1	X1	6
2	1	X2	1
2	2	X1	2
2	2	X2	4

长表转为表 e

md |> 
  pivot_wider(
    names_from = Time,
    names_prefix = "Time",
    values_from = value
  ) -> table_e
kable(table_e, align = "c")

ID	variable	Time1	Time2
1	X1	5	3
1	X2	6	5
2	X1	6	2
2	X2	1	4

表 e 逆回长表

table_e |> 
  pivot_longer(
    cols = starts_with("Time"), 
    names_to = "Time",
    names_transform = list(Time = readr::parse_number),
    values_to = "value"
  ) |> 
  kable(align = "c")

ID	variable	Time	value
1	X1	1	5
1	X1	2	3
1	X2	1	6
1	X2	2	5
2	X1	1	6
2	X1	2	2
2	X2	1	1
2	X2	2	4

长表转为表 f

md |> 
  pivot_wider(
    names_from = c(variable, Time), 
    names_glue = "{variable}_Time{Time}",
    values_from = value
  ) -> table_f
kable(table_f, align = "c")

ID	X1_Time1	X2_Time1	X1_Time2	X2_Time2
1	5	6	3	5
2	6	1	2	4

表 f 逆回长表

table_f |> 
  pivot_longer(
    -ID,
    names_to = c("variable", "Time"),
    names_sep = "_Time",
    names_transform = list(Time = as.numeric),
    values_to = "value"
  ) |> 
  kable(align = "c")

ID	variable	Time	value
1	X1	1	5
1	X2	1	6
1	X1	2	3
1	X2	2	5
2	X1	1	6
2	X2	1	1
2	X1	2	2
2	X2	2	4