Question

我的数据

Chemical  date      concentration  limit
A     01-01-2016     0.2         0.01
A     01-02-2016     0.2         0.01
A     01-01-2017     0.005       0.01
A     01-02-2017     0.2         0.01
B     01-01-2016     0.3         0.1
B     01-02-2016     0.05        0.1
B     01-01-2017     0.2         0.1
B     01-02-2017     0.2         0.1
C     01-01-2016     1.2         1
C     01-02-2016     0.8         1
C     01-01-2017     0.9         1
C     01-02-2017     0.9         1

我想显示每种化学物质超过其每年限量的百分比（请注意，每种限量都不同）。所以我想得到这样的东西

Year   A         B     C
2016   100%     50%   50%
2017   50%      100%   0

我已经有了代码来计算每种化学物质每年超过的次数，但是当计算百分比时，我会弄错了。

这我要数一遍。

  library(tidyverse)
   counts<- data %>% 
   group_by(Chemical, grp = format(date, format = '%Y')) %>% 
   mutate(exceed = concentration >= limit) %>% # TRUE/FALSE
   summarise(tot_exceed = sum(exceed)) %>%  # count each T/F
   spread(Chemical, tot_exceed, fill = 0)

所以我明白了

   Year   A     B    C
   2016   2     1    1
   2017   1     2    0

对于百分比，我尝试了。

percentage_exceed<- data %>% 
group_by(Chemical, grp = format(date, format = '%Y')) %>% 
mutate(exceed = concentration >= limit, countconc = length(concentration)) 
%>% 
summarise(percent = (sum(exceed)/countconc)*100) %>% 
spread(Chemical, percent, fill = 0)

但是我没有得到想要的结果。你能帮我吗？

Answer 1

使用tidyverse：

library(tidyverse)
library(lubridate)

 data  %>% 
   mutate(yr=mdy(date) %>% year) %>% 
   group_by(Chemical,yr) %>% 
   mutate(exceed  = ifelse(concentration>=limit,1,0  )) %>% 
   summarise(tot_exceed =sum(exceed)) %>% 
   group_by(Chemical) %>% 
   mutate(proc=tot_exceed/max(tot_exceed)*100) %>% 
   select(-tot_exceed) %>% 
   spread(Chemical,proc)
# A tibble: 2 x 4
     yr     A     B     C
  <dbl> <dbl> <dbl> <dbl>
1  2016   100    50   100
2  2017    50   100     0

Answer 2

dt = read.table(text = "
Chemical  date      concentration  limit
A     01-01-2016     0.2         0.01
A     01-02-2016     0.2         0.01
A     01-01-2017     0.005       0.01
A     01-02-2017     0.2         0.01
B     01-01-2016     0.3         0.1
B     01-02-2016     0.05        0.1
B     01-01-2017     0.2         0.1
B     01-02-2017     0.2         0.1
C     01-01-2016     1.2         1
C     01-02-2016     0.8         1
C     01-01-2017     0.9         1
C     01-02-2017     0.9         1
", header=T)

library(tidyverse)
library(lubridate)

dt %>%
  mutate(year = year(dmy(date))) %>%
  group_by(year, Chemical) %>%
  summarise(Total = n(),
            Num_exceed = sum(concentration >= limit)) %>%
  ungroup() %>%
  mutate(Prc = paste0(Num_exceed / Total * 100,"%")) %>%
  select(year, Chemical, Prc) %>%
  spread(Chemical, Prc)

# # A tibble: 2 x 4
#    year A     B     C    
#   <dbl> <chr> <chr> <chr>
# 1  2016 100%  50%   50%  
# 2  2017 50%   100%  0%

Answer 3

您的方法非常好，您只需将sum替换为mean，然后乘以100：

data %>% group_by(Chemical, grp = format(date, format = '%Y')) %>% 
  mutate(exceed = concentration >= limit) %>% 
  summarise(tot_exceed = mean(exceed) * 100) %>%
  spread(Chemical, tot_exceed, fill = 0)
# A tibble: 2 x 4
#   grp       A     B     C
#   <chr> <dbl> <dbl> <dbl>
# 1 2016    100    50    50
# 2 2017     50   100     0

您尝试的线路

summarise(percent = (sum(exceed)/countconc) * 100)

几乎可以做到这一点：错误是因为countconc是整列而不是单个值（汇总所需）。因此，由于无论如何它都是每个组中的恒定列，因此您可以编写例如

summarise(percent = (sum(exceed)/countconc[1]) * 100)

但是在前面的那一行，

mutate(exceed = concentration >= limit, countconc = length(concentration))

这最终只是一个意思，所以我们回到答案开头的代码。

还要注意，使用lubridate可以将第一行写为

data %>% group_by(Chemical, Year = year(date)) %>%

某些东西非常简洁，但可能不是您想要的格式

data %>% group_by(Chemical, Year = year(date)) %>% 
  summarise(Percentage = mean(concentration > limit) * 100)
# A tibble: 6 x 3
# Groups:   Chemical [?]
#   Chemical  Year Percentage
#   <fct>    <dbl>      <dbl>
# 1 A         2016        100
# 2 A         2017         50
# 3 B         2016         50
# 4 B         2017        100
# 5 C         2016         50
# 6 C         2017          0

Answer 4

使用tidyverse和reshape2，您可以执行以下操作：

df %>%
 mutate(date = str_sub(as.character(date), 7, 10)) %>% #Taking out the year from "date"
 group_by(date, Chemical) %>% #Group by "date" and "Chemical"
 summarise(temp = length(concentration[concentration > limit])/n()*100) %>% #Applying the condition
 dcast(date~Chemical, value.var = "temp") #Reshaping the data

  date   A   B  C
1 2016 100  50 50
2 2017  50 100  0

或者仅使用tidyverse的{{1}}

spread()

每年每个值的百分比

4 个答案: