3

我有一个数据框，其列具有多个值（逗号分隔）：

mydf <- structure(list(Age = c(99L, 10L, 40L, 15L),
                       Info = c("good, bad, sad", "nice, happy, joy", "NULL", "okay, nice, fun, wild, go"),
                       Target = c("Boy", "Girl", "Boy", "Boy")), 
                  .Names = c("Age", "Info", "Target"),
                  row.names = c(NA, 4L),
                  class = "data.frame")

> mydf
  Age                      Info Target
1  99            good, bad, sad    Boy
2  10          nice, happy, joy   Girl
3  40                      NULL    Boy
4  15 okay, nice, fun, wild, go    Boy

我想将Info列拆分为one-hot-encoded列，并将结果附加到 Target 列之外，例如：

  Age                      Info Target good bad sad nice ... NULL ..
1  99            good, bad, sad    Boy    1   1   1   0        0
2  10          nice, happy, joy   Girl    0   0   0   1        0
3  40                      NULL    Boy    0   0   0   0        1
4  15 okay, nice, fun, wild, go    Boy    0   0   0   0        0

在 python 中，我可以执行以下操作来获取字典，然后使用它来分配列。

In [1]: import itertools

In [2]: values = ["good, bad, sad", "nice, happy, joy", "NULL",  "okay, nice, fun, wild, go"]

In [3]: terms = list(itertools.chain(*[v.split(", ") for v in values]))

In [4]: dictionary = {v:k for k,v in enumerate(terms)}

In [6]: dictionary
Out[6]: 
{'NULL': 6, 'bad': 1,
 'fun': 9, 'go': 11, 'good': 0, 'happy': 4,
 'joy': 5, 'nice': 8, 'okay': 7, 'sad': 2, 'wild': 10}

到目前为止，我可以在 R 中做到这一点

> lapply(mydf["Info"], function(x) { strsplit(x, ", ") } )
$Info
$Info[[1]]
[1] "good" "bad"  "sad" 

$Info[[2]]
[1] "nice"  "happy" "joy"  

$Info[[3]]
[1] "NULL"

$Info[[4]]
[1] "okay" "nice" "fun"  "wild" "go"

我不知道如何将其转换为 R 中的字典，并使用它转换为 One-Hot-Encoding 的列。

我该如何解决这个问题？

4

1 回答 1

10

一种选择是mtabulate在qdapTools将“信息”列拆分为,

library(qdapTools)
cbind(mydf, mtabulate(strsplit(mydf$Info, ", ")))
#Age                      Info Target bad fun go good happy joy nice NULL okay sad wild
#1  99            good, bad, sad    Boy   1   0  0    1     0   0    0    0    0   1    0
#2  10          nice, happy, joy   Girl   0   0  0    0     1   1    1    0    0   0    0
#3  40                      NULL    Boy   0   0  0    0     0   0    0    1    0   0    0
#4  15 okay, nice, fun, wild, go    Boy   0   1  1    0     0   0    1    0    1   0    1
于 2016-09-29T19:24:24.340 回答