无法使用 dplyr mutate 添加列答案

【问题标题】：unable to add column with dplyr mutate无法使用 dplyr mutate 添加列
【发布时间】：2017-02-01 19:01:21
【问题描述】：

我目前有这个数据框

    dat = data.frame(time= c("Q1","Q2","Q3"), 
                     measure1 = c(1,2,9   ),
                     measure2 = c(4,5,6   ))
    dat2 = dat %>% gather(key= Metric, value = Value, c(measure1, measure2)) %>% group_by(Metric, time)
dat2 = as.data.frame(dat2)

time   Metric Value
1   Q1 measure1     1
2   Q2 measure1     2
3   Q3 measure1     9
4   Q1 measure2     4
5   Q2 measure2     5
6   Q3 measure2     6

我可以像这样添加这个“测试”列

dat2$test=  ifelse(  dat2$Metric =="measure1" &  dat2$Value >   dat2$Value[ dat2$Metric=="measure2"] ,1,
                               ifelse(  dat2$Metric == "measure2" &   dat2$Value >  dat2$Value[ dat2$Metric=="measure1"] ,1,-1)
                               ) #end ifelse

 time   Metric Value test
1   Q1 measure1     1   -1
2   Q2 measure1     2   -1
3   Q3 measure1     9    1
4   Q1 measure2     4    1
5   Q2 measure2     5    1
6   Q3 measure2     6   -1

但我想像这样使用 dplyr 和变异

dat = data.frame(time= c("Q1","Q2","Q3"), 
                 measure1 = c(1,2,9   ),
                 measure2 = c(4,5,6   ))
dat %>% gather(key= Metric, value = Value, c(measure1, measure2)) %>% group_by(Metric, time) %>%
     mutate(test= ifelse( Metric =="measure1" &  Value > Value[Metric=="measure2"] ,1,
                               ifelse( Metric =="measure2" &  Value > Value[Metric=="measure1"] ,1,-1)
                               ) #end ifelse
                        )#end mutate
dat2 = as.data.frame(dat2)
dat2

但“测试”列全是 NA

 time   Metric Value test
1   Q1 measure1     1   NA
2   Q2 measure1     2   NA
3   Q3 measure1     9   NA
4   Q1 measure2     4   NA
5   Q2 measure2     5   NA
6   Q3 measure2     6   NA

为什么不能使用 mutate 添加列？是否与使用 as.data.frame 并手动添加有关...mutate 无法识别列名？

谢谢。

【问题讨论】：

标签： r dplyr

【解决方案1】：

问题在于，当您按Metric 分组时，对于Metric 为measure1、Metric == "measure2" 将始终返回 FALSE 和Value[FALSE] == numeric(0) 的组，零长度对象在之后转换为 NA mutate:

dat %>% 
        gather(key= Metric, value = Value, c(measure1, measure2)) %>% 
        group_by(time) %>%        # <<<<<<<<<<<< here
        mutate(test= ifelse(Metric =="measure1" & Value > Value[Metric=="measure2"] ,1,
                            ifelse(Metric =="measure2" & Value > Value[Metric=="measure1"],1,-1)
                            ) #end ifelse
              )

# Source: local data frame [6 x 4]
# Groups: time [3]

#     time   Metric Value  test
#   <fctr>    <chr> <dbl> <dbl>
# 1     Q1 measure1     1    -1
# 2     Q2 measure1     2    -1
# 3     Q3 measure1     9     1
# 4     Q1 measure2     4     1
# 5     Q2 measure2     5     1
# 6     Q3 measure2     6    -1

通过此示例按指标分组时，您可以更清楚地看到发生了什么：

dat %>% 
        gather(key= Metric, value = Value, c(measure1, measure2)) %>% 
        group_by(Metric, time) %>%
        mutate(test= Value[Metric == "measure2"])

# Source: local data frame [6 x 4]
# Groups: Metric, time [6]

#     time   Metric Value  test
#   <fctr>    <chr> <dbl> <dbl>
# 1     Q1 measure1     1    NA
# 2     Q2 measure1     2    NA
# 3     Q3 measure1     9    NA
# 4     Q1 measure2     4     4
# 5     Q2 measure2     5     5
# 6     Q3 measure2     6     6

【讨论】：