创建“词典/参考”桌子

Question

我有此数据集：

col_1 = as.factor(c("a", "a", "b", "c", "b", "a"))
col_2 = c(15, 346, 3564, 99, 10, 2)
col_3 = as.factor(c("bb", "a", "g", "f", "bb", "a"))
index = 1:6

sample_data = data.frame(index, col_1, col_2, col_3)

  index col_1 col_2 col_3
     1     1    15     4
     2     1   346     5
     3     2  3564     6
     4     3    99     7
     5     2    10     4
     6     1     2     5

在另一个问题中（ sequentallylylylylylylylylylylylylylyllylylylylylylylyly innumerical actical variables具有数值），我学会了如何用数字列举所有因素变量：

indx <- vapply(sample_data, is.factor, logical(1))
vec <- interaction(stack(type.convert(sample_data[,indx], as.is = TRUE)))
sample_data[indx] <- match(vec, unique(vec))

  index col_1 col_2 col_3
1     1     1    15     4
2     2     1   346     5
3     3     2  3564     6
4     4     3    99     7
5     5     2    10     4
6     6     1     2     5

我想尝试创建一个“字典表”（即“传奇”）显示了原始数据与转换数据之间的关系。我想出了一种手动执行此操作的方法：

library(plyr)

col_1_legend = unique(data.frame(original_data$col_1, sample_data$col_1))
col_3_legend = unique(data.frame(original_data$col_3, sample_data$col_3))

dictionary_data<- plyr::rbind.fill(col_1_legend,col_3_legend)

 original_data.col_1 sample_data.col_1 original_data.col_3 sample_data.col_3
1                   a                 1                <NA>                NA
2                   b                 2                <NA>                NA
3                   c                 3                <NA>                NA
4                <NA>                NA                  bb                 4
5                <NA>                NA                   a                 5
6                <NA>                NA                   g                 6
7                <NA>                NA                   f                 7

但是，这是一种创建“字典表”的非常混乱且效率低下的方法（例如，有很多带有因子变量的列？）。有人可以建议一种更有效的方法吗？

谢谢你！

Answer 1

我会如下创建它； ）的数据

和match （

# sample 1
col_1 = as.factor(c("a", "a", "b", "c", "b", "a"))
col_2 = c(15, 346, 3564, 99, 10, 2)
col_3 = as.factor(c("bb", "a", "g", "f", "bb", "a"))
index = 1:6

sample_df1 = data.frame(index, col_1, col_2, col_3)

# sample 2
indx <- vapply(sample_df1, is.factor, logical(1))

stacked <- stack(type.convert(sample_df1[,indx], as.is = TRUE))
vec <- interaction(stacked)
matched <- match(vec, unique(vec))

sample_df2 <- sample_df1
sample_df2[indx] <- matched

作为长格式数据框架，带有来自stack（）

# create legend
legend <- cbind(matched, stacked) %>% distinct()
legend <- legend[c(3,2,1)]
colnames(legend) <- c('column', 'original_data', 'sample_data')

> legend
  column original_data sample_data
1  col_1             a           1
2  col_1             b           2
3  col_1             c           3
4  col_3            bb           4
5  col_3             a           5
6  col_3             g           6
7  col_3             f           7