Question

我有一个具有以下结构的数据框：

library(tidyverse)

df <- tibble::tribble(
              ~var1,            ~var2,              ~var3, 
        "ano 2005",      "km 128000",         "marca chevrolet",
        "ano 2019",     "marca hyundai tucson",        "km 50000",
    "marca grand vitara sz",     "ano 2012",              "NA"
    )

我需要使用以下代码创建新变量，并为此分配相应的信息

df %>% 
stack() %>% 
select(-ind) %>% 
separate(values, into = c("column", "value")) %>% 
pivot_wider(value, column, values_fn = list(value = list)) %>% 
unnest(cols = c(marca, ano, km))

但不适用，我遇到以下错误：marca，大小为120，km，大小为119，没有通用大小。还有另一个错误，它仅返回第一个单词，其余的单词将其消除。如果有人可以帮助我，我将非常感谢

 ano   marca        km
2005   chevrolet  128000
2019   hyundai     50000
2012   grand

Answer 1

这是我的data.table方法

library( data.table )

#set to data.table format
setDT(df)
#create row_id's
df[, id := .I][]
#melt to long
ans <- melt( df, id.vars = "id" )
#split strings, using first space as separator
ans[, c("col_name", "col_value") := as.data.table( stringr::str_split_fixed( value, " ", 2 ) ) ]
#cast to wide
dcast( ans[!col_name == "NA",], id ~ col_name, value.var = "col_value")

#    id  ano     km           marca
# 1:  1 2005 128000       chevrolet
# 2:  2 2019  50000  hyundai tucson
# 3:  3 2012   <NA> grand vitara sz

Answer 2

这是一个tidyverse解决方案。这要求数据框中的NA值为NA_character_，而不是带引号的字符串"NA"。参数extra='merge'告诉separate()不要在汽车型号名称中添加其他单词。

# replace quoted NA with true NA
df[df == 'NA'] <- NA_character_

 df %>%
  mutate(id = 1:nrow(df)) %>%
  pivot_longer(-id) %>%
  separate(value, into = c('column', 'value'), extra = 'merge') %>%
  select(-name) %>%
  filter(!is.na(column)) %>%
  pivot_wider(id_cols = id, names_from = column)

# A tibble: 3 x 4
#     id ano   km     marca          
#  <int> <chr> <chr>  <chr>          
#1     1 2005  128000 chevrolet      
#2     2 2019  50000  hyundai tucson 
#3     3 2012  NA     grand vitara sz

使用ivot_wider从杂乱的数据中创建新变量，这些杂乱的数据在多列中具有变量

2 个答案: