R循环遍历数据框列中的唯一值以根据条件创建另一个

Question

My dataset consists of scores and total respondents for questions asked in a survey, over a number of fiscal years (FY13, FY14 & FY15) and in different regions.

My objective is to loop through the FY column and identify when each question was asked, for each region. And store this information in a new column.

This is what a reproducible sample looks like -

testdf=data.frame(FY=c("FY13","FY14","FY15","FY14","FY15","FY13","FY14","FY15","FY13","FY15","FY13","FY14","FY15","FY13","FY14","FY15"),
              Region=c(rep("AFRICA",5),rep("ASIA",5),rep("AMERICA",6)),
              QST=c(rep("Q2",3),rep("Q5",2),rep("Q2",3),rep("Q5",2),rep("Q2",3),rep("Q5",3)),
              Very.Satisfied=runif(16,min = 0, max=1),
              Total.Very.Satisfied=floor(runif(16,min=10,max=120)),
              Satisfied=runif(16,min = 0, max=1),
              Total.Satisfied=floor(runif(16,min=10,max=120)),
              Dissatisfied=runif(16,min = 0, max=1),
              Total.Dissatisfied=floor(runif(16,min=10,max=120)),
              Very.Dissatisfied=runif(16,min = 0, max=1),
              Total.Very.Dissatisfied=floor(runif(16,min=10,max=120)))

I start with creating an ID column, by concatenating Region & QST

library(tidyr)
testdf = testdf %>%
unite(ID,c('Region','QST'),sep = "",remove = F)

My Objective

1) For each unique ID , identify whether the given question was asked -

a) Only on one year (either FY13, FY14 or FY15)

b) Over the Past Two Years (FY15 & FY14 only)

c) Over the Past Three Years (FY15 & FY14 & FY13)

d) On FY13 & FY15 Only

My Attempt

For this problem, I tried to create a for loop , and for each unique ID , I first store the unique occurences of each FY the question was asked in a vector v . Then using an IF conditional statement I assign a comment to a newly created column called Tally based on these occurences.

for (i in unique(testdf$ID))
{
v=unique(testdf$FY)

  if(('FY15' %in% v) & ('FY14' %in% v)) {
      testdf$Tally=='Asked Over The Past Two Years'
  } 
  else if(('FY15' %in% v) & ('FY14' %in% v) & ('FY13' %in% v)) {
       testdf$Tally=='Asked Over The Past Three Years'
  }
  else if(('FY13' %in% v) & ('FY15' %in% v)) {
        testdf$Tally=='Question Asked in FY13 & FY15 Only'
  }
  else { testdf$Tally=='Question Asked Once Only' 
  }

}  

The loop seems to run without throwing an error message, but it doesn't seem to create the new Tally column.

Any help with this will be greatly appreciated.

Answer 1

In your code the main problem is that in the if-else clause you're not doing an assignment (using '<-') but a comparison, using '=='. Here's a solution that I find more elegant, since it's not using a loop:

require(tidyverse)

testdf %>%
  select(ID, FY) %>%
  unique() %>%
  mutate(is_true = 1) %>%
  spread(key = FY, value = is_true, fill = 0) %>%
  mutate(tally = case_when(
    FY13 == 1 & FY14 == 1 & FY15 == 1 ~ 'Asked Over The Past Three Years',
                FY14 == 1 & FY15 == 1 ~ 'Asked Over the Past Two Years',
    FY13 == 1 &             FY15 == 1 ~ 'Asked in FY12 & FY15 Only',
    TRUE ~ 'Question Asked Once Only'
  ))

Output:

+------------------------------------------------------------+
|          ID FY13 FY14 FY15                           tally |
+------------------------------------------------------------+
| 1  AFRICAQ2    1    1    1 Asked Over The Past Three Years |
| 2  AFRICAQ5    0    1    1   Asked Over the Past Two Years |
| 3 AMERICAQ2    1    1    1 Asked Over The Past Three Years |
| 4 AMERICAQ5    1    1    1 Asked Over The Past Three Years |
| 5    ASIAQ2    1    1    1 Asked Over The Past Three Years |
| 6    ASIAQ5    1    0    1       Asked in FY12 & FY15 Only |
+------------------------------------------------------------+

Answer 2

No need for a loop:

library(tidyverse)

result <- testdf %>%
    select(3, 2, 1) %>%
    mutate(Asked = 1) %>%
    spread(FY, Asked)

> result
  QST  Region FY13 FY14 FY15
1  Q2  AFRICA    1    1    1
2  Q2 AMERICA    1    1    1
3  Q2    ASIA    1    1    1
4  Q5  AFRICA   NA    1    1
5  Q5 AMERICA    1    1    1
6  Q5    ASIA    1   NA    1

Answers all four questions in one go.

If you really want a tally column, expand it like this:

result %>%
    mutate(Tally = case_when(FY13 + FY14 + FY15 == 1 ~ "Only one year",
                             FY13 + FY14 + FY15 == 3 ~ "Past three years",
                             FY14 + FY15 == 2 ~ "Past two years",
                             FY13 + FY15 == 2 ~ "FY13 and FY15 only",
                             NA ~ NA_character_))

  QST  Region FY13 FY14 FY15              Tally
1  Q2  AFRICA    1    1    1   Past three years
2  Q2 AMERICA    1    1    1   Past three years
3  Q2    ASIA    1    1    1   Past three years
4  Q5  AFRICA   NA    1    1     Past two years
5  Q5 AMERICA    1    1    1   Past three years
6  Q5    ASIA    1   NA    1 FY13 and FY15 only

Answer 3

Consider ave for grouping calculation by Region and QST inside nested ifelse for conditional logic:

testdf <- within(testdf, {
                   FY13 <- ifelse(FY=='FY13', 1, 0)
                   FY14 <- ifelse(FY=='FY14', 1, 0)
                   FY15 <- ifelse(FY=='FY15', 1, 0)

                   Tally <- ifelse(ave(FY13, Region, QST, FUN=max) + ave(FY14, Region, QST, FUN=max) + ave(FY15, Region, QST, FUN=max) == 1,
                                   'Asked Only on One Year',
                                   ifelse(ave(FY13, Region, QST, FUN=max) + ave(FY14, Region, QST, FUN=max) + ave(FY15, Region, QST, FUN=max) == 3,
                                          'Asked Over the Past Three Years',
                                          ifelse(ave(FY14, Region, QST, FUN=max) + ave(FY15, Region, QST, FUN=max) == 2,
                                                 'Asked Over the Past Two Years',
                                                 ifelse(ave(FY13, Region, QST, FUN=max) + ave(FY15, Region, QST, FUN=max) == 2,
                                                        'Asked On FY13 & FY15 Only',
                                                        NA
                                                        )
                                                 )
                                          )
                                   )

                   FY13 <- NULL; FY14 <- NULL; FY15 <- NULL
             })

testdf[c("ID", "FY", "Tally")]

#     Region QST   FY                           Tally
# 1   AFRICA  Q2 FY13 Asked Over the Past Three Years
# 2   AFRICA  Q2 FY14 Asked Over the Past Three Years
# 3   AFRICA  Q2 FY15 Asked Over the Past Three Years
# 4   AFRICA  Q5 FY14   Asked Over the Past Two Years
# 5   AFRICA  Q5 FY15   Asked Over the Past Two Years
# 6     ASIA  Q2 FY13 Asked Over the Past Three Years
# 7     ASIA  Q2 FY14 Asked Over the Past Three Years
# 8     ASIA  Q2 FY15 Asked Over the Past Three Years
# 9     ASIA  Q5 FY13       Asked On FY13 & FY15 Only
# 10    ASIA  Q5 FY15       Asked On FY13 & FY15 Only
# 11 AMERICA  Q2 FY13 Asked Over the Past Three Years
# 12 AMERICA  Q2 FY14 Asked Over the Past Three Years
# 13 AMERICA  Q2 FY15 Asked Over the Past Three Years
# 14 AMERICA  Q5 FY13 Asked Over the Past Three Years
# 15 AMERICA  Q5 FY14 Asked Over the Past Three Years
# 16 AMERICA  Q5 FY15 Asked Over the Past Three Years

Answer 4

There's a solution using your ID column. (Using paste0 we can do that somewhat nicer, though with testdf$ID <- paste0(testdf$Region, "_", testdf$QST) .)

We dcast your testdf using the reshape2 package.

library(reshape2)
tmp <- dcast(testdf, ID ~ FY, 
               value.var="QST", fun.aggregate=length)

Now we already know whether the question was asked in the different years. To answer the further questions, we'll do some maths.

tmp <- cbind(tmp, 
             past2=as.numeric(t2[3] + t2[4] == 2 & t2[2] == 0), 
             past3=as.numeric(t2[2] + t2[3] + t2[4] == 3),
             y13_15=as.numeric(t2[2] + t2[4] == 2 & t2[3] == 0))

The sequences in the 5:7 columns contain the desired Tally information that we can milk

tmp$Tally <- apply(tmp, 1, function(x) paste0(x[5:7], collapse=""))

translate into human language by factor levels,

tmp$Tally <- factor(tmp$Tally, labels=c('Question Asked Once Only',
                                        'Question Asked in FY13 & FY15 Only',
                                        'Asked Over The Past Three Years',
                                        'Asked Over The Past Two Years'))

and merge with the original data frame to achieve the desired result.

Result

> merge(testdf, t3[c(1, 8)])
             ID   FY    Region QST                              Tally
1     AFRICA_Q2 FY13    AFRICA  Q2    Asked Over The Past Three Years
2     AFRICA_Q2 FY14    AFRICA  Q2    Asked Over The Past Three Years
3     AFRICA_Q2 FY15    AFRICA  Q2    Asked Over The Past Three Years
4     AFRICA_Q5 FY14    AFRICA  Q5      Asked Over The Past Two Years
5     AFRICA_Q5 FY15    AFRICA  Q5      Asked Over The Past Two Years
6    AMERICA_Q2 FY13   AMERICA  Q2    Asked Over The Past Three Years
7    AMERICA_Q2 FY14   AMERICA  Q2    Asked Over The Past Three Years
8    AMERICA_Q2 FY15   AMERICA  Q2    Asked Over The Past Three Years
9    AMERICA_Q5 FY13   AMERICA  Q5    Asked Over The Past Three Years
10   AMERICA_Q5 FY14   AMERICA  Q5    Asked Over The Past Three Years
11   AMERICA_Q5 FY15   AMERICA  Q5    Asked Over The Past Three Years
12 ANTH.CTRY_Q2 FY15 ANTH.CTRY  Q2           Question Asked Once Only
13      ASIA_Q2 FY13      ASIA  Q2    Asked Over The Past Three Years
14      ASIA_Q2 FY14      ASIA  Q2    Asked Over The Past Three Years
15      ASIA_Q2 FY15      ASIA  Q2    Asked Over The Past Three Years
16      ASIA_Q5 FY13      ASIA  Q5 Question Asked in FY13 & FY15 Only
17      ASIA_Q5 FY15      ASIA  Q5 Question Asked in FY13 & FY15 Only

---

Data

testdf <- structure(list(FY = c("FY13", "FY14", "FY15", "FY14", "FY15", 
"FY13", "FY14", "FY15", "FY13", "FY15", "FY13", "FY14", "FY15", 
"FY13", "FY14", "FY15", "FY15"), Region = c("AFRICA", "AFRICA", 
"AFRICA", "AFRICA", "AFRICA", "ASIA", "ASIA", "ASIA", "ASIA", 
"ASIA", "AMERICA", "AMERICA", "AMERICA", "AMERICA", "AMERICA", 
"AMERICA", "ANTH.CTRY"), QST = c("Q2", "Q2", "Q2", "Q5", "Q5", 
"Q2", "Q2", "Q2", "Q5", "Q5", "Q2", "Q2", "Q2", "Q5", "Q5", "Q5", 
"Q2")), row.names = c(NA, 17L), class = "data.frame")