从长表创建二进制宽表（如tidyr :: spread（））

Question

作为树模型的输入，我在SQL中创建了一个分析表。现在，我想将其传输到R，因为以该表为输入的模型也在R中运行。我无法转换为R的SQL步骤之一。

分析表具有以下形式：

df <- data.frame(
  pseudonym = c("a", "a", "a", "b", "c", "c"),
  var1 = c(1,1,0,1,1,0),
  var2 = c(1,0,0,0,0,1),
  var3 = c(0,0,0,0,0,1))

> df
  pseudonym var1 var2 var3
1         a    1    1    0
2         a    1    0    0
3         a    0    0    0
4         b    1    0    0
5         c    1    0    0
6         c    0    1    1

在下一步中，我需要假名的不同行，同时保留其他列var1，var2，var3中的信息（1）。 （在SQL中，这是通过max(case when...then 1 else 0 end) as var1）

因此从df1创建的结果df2应该是

df2 <- data.frame(
  pseudonym = c("a", "b", "c"),
  var1 = c(1,1,1),
  var2 = c(1,0,1),
  var3 = c(0,0,1))

> df2
  pseudonym var1 var2 var3
1         a    1    1    0
2         b    1    0    0
3         c    1    1    1

如果有人有一个主意，那将非常有帮助。

Answer 1

这里是一种方式：

library(dplyr)
library(tidyr)

df <- data.frame(
  pseudonym = c("a", "a", "a", "b", "c", "c"),
  var1 = c(1,1,0,1,1,0),
  var2 = c(1,0,0,0,0,1),
  var3 = c(0,0,0,0,0,1))

df %>% 
  pivot_longer(cols = var1:var3) %>% 
  group_by(pseudonym, name) %>% 
  filter(max(value) == value) %>% 
  ungroup() %>% 
  distinct() %>% 
  pivot_wider(names_from = name, values_from = value)

#># A tibble: 3 x 4
#>  pseudonym  var1  var2  var3
#>  <fct>     <dbl> <dbl> <dbl>
#>1 a             1     1     0
#>2 b             1     0     0
#>3 c             1     1     1

Answer 2

另一种dplyr方法，可能不太复杂，但可以起作用：

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
df <- data.frame(
    pseudonym = c("a", "a", "a", "b", "c", "c"),
    var1 = c(1,1,0,1,1,0),
    var2 = c(1,0,0,0,0,1),
    var3 = c(0,0,0,0,0,1)); df
#>   pseudonym var1 var2 var3
#> 1         a    1    1    0
#> 2         a    1    0    0
#> 3         a    0    0    0
#> 4         b    1    0    0
#> 5         c    1    0    0
#> 6         c    0    1    1

df2 <- df %>% group_by(pseudonym) %>% mutate(var1 = case_when(1 %in% var1 ~ 1),
                                      var2 = case_when(1 %in% var2 ~ 1),
                                      var3 = case_when(1 %in% var3 ~ 1)) %>% 
                                      unique() %>% 
    ungroup() #creates "NA" in place of "0" 
df2[is.na(df2)] <- 0; df2 #"NA"s are taken care of
#> # A tibble: 3 x 4
#>   pseudonym  var1  var2  var3
#>   <fct>     <dbl> <dbl> <dbl>
#> 1 a             1     1     0
#> 2 b             1     0     0
#> 3 c             1     1     1

^{由reprex package（v0.3.0）在2020-04-21创建}