从 R 中具有不同行长度的数据框创建关联矩阵

Question

我有一个 space 分隔文件，类似于：

# 1_1 AAA ABA AAB BBA
# 1_2 ABA AAA ABB BAA
# 1_3 ABA BAA
# 1_4 AAA BBB ABB

但是，行的内容（我们称它们为“单词”）包含更多字符（大约 20 个）。每行中的单词数各不相同（从少于 10 个到几十个）。我想使用 R 创建一个关联矩阵，其中每个唯一的单词都是一列，行名称将保留为行名称，如下所示：

#     AAA AAB ABA ABB BAA BBA BBB
# 1_1  1   1   1   0   0   1   0
# 1_2  1   0   1   1   1   0   0
# 1_3  0   0   1   0   1   0   0
# 1_4  1   0   0   1   0   0   1

列顺序不相关。非常感谢任何建议，欢迎使用基础解决方案和包解决方案。谢谢！

Answer 1

tidyverse

library(tidyverse)

df <-
  structure(
    list(
      id = c("1_1", "1_2", "1_3", "1_4"),
      col_1 = c("AAA", "ABA", "ABA", "AAA"),
      col_2 = c("ABA", "AAA", "BAA", "BBB"),
      col_3 = c("AAB", "ABB", NA, "ABB"),
      col_4 = c("BBA", "BAA", NA, NA)
    ),
    class = "data.frame",
    row.names = c(NA,-4L)
  )

df %>%
  pivot_longer(-id, values_drop_na = TRUE) %>%
  pivot_wider(
    id_cols = id,
    names_from = value,
    values_from = value,
    values_fn = length,
    values_fill = 0
  )
#> # A tibble: 4 x 8
#>   id      AAA   ABA   AAB   BBA   ABB   BAA   BBB
#>   <chr> <int> <int> <int> <int> <int> <int> <int>
#> 1 1_1       1     1     1     1     0     0     0
#> 2 1_2       1     1     0     0     1     1     0
#> 3 1_3       0     1     0     0     0     1     0
#> 4 1_4       1     0     0     0     1     0     1

^{由 reprex package (v2.0.1)}

于 2021-12-13 创建

data.table

library(data.table)
library(magrittr)

setDT(df)

melt(df, id.vars = "id") %>% 
  na.omit() %>% 
  dcast(formula = id ~ value, fun.aggregate = length)
#>     id AAA AAB ABA ABB BAA BBA BBB
#> 1: 1_1   1   1   1   0   0   1   0
#> 2: 1_2   1   0   1   1   1   0   0
#> 3: 1_3   0   0   1   0   1   0   0
#> 4: 1_4   1   0   0   1   0   0   1

^{由 reprex package (v2.0.1)}

于 2021-12-13 创建

基础

df_stack <- stack(df, select = -id)
df_stack$id <- rep(df$id, ncol(df) - 1)
+(with(df_stack, table(id, values)))
#>      values
#> id    AAA AAB ABA ABB BAA BBA BBB
#>   1_1   1   1   1   0   0   1   0
#>   1_2   1   0   1   1   1   0   0
#>   1_3   0   0   1   0   1   0   0
#>   1_4   1   0   0   1   0   0   1

^{由 reprex package (v2.0.1)}

于 2021-12-13 创建

Answer 2

试试下面的代码

> +(t(table(stack(as.data.frame(t(`row.names<-`(df[-1], df$id)))))) > 0)
     values
ind   AAA AAB ABA ABB BAA BBA BBB
  1_1   1   1   1   0   0   1   0
  1_2   1   0   1   1   1   0   0
  1_3   0   0   1   0   1   0   0
  1_4   1   0   0   1   0   0   1

Answer 3

选项unite/cSplit_e

library(dplyr)
library(tidyr)
library(splitstackshape)
library(stringr)
df %>% 
  unite(new, starts_with('col'), na.rm = TRUE) %>% 
  cSplit_e(., "new", sep="_", type = "character", fill = 0, drop = TRUE) %>% 
  rename_with(~ str_remove(., "new_"), -id)
   id AAA AAB ABA ABB BAA BBA BBB
1 1_1   1   1   1   0   0   1   0
2 1_2   1   0   1   1   1   0   0
3 1_3   0   0   1   0   1   0   0
4 1_4   1   0   0   1   0   0   1

从 R 中具有不同行长度的数据框创建关联矩阵

Creating an incidence matrix from data frame with distinct row lengths in R

r

matrix