我正在使用excel电子表格中的一些活动前和活动后数据，其中一个问题可能有多个答案（并且学生必须选择所有正确的答案以获得该问题的完全信用），但数据的方式最初格式化，每个单独的响应列在自己的行中，其他元数据（数据库信息，学生用户名，性别等）只是为每一行复制 - 我将通过下面的示例详细说明。
>data
Database Username Gender InterviewType Question Answer
1 steve Male Pre Which of the following... "Response 1"
1 steve Male Pre Which of the following... "Response 2"
1 steve Male Pre Which of the following... "Response 3"
1 steve Male Pre Please indicate your race. "White"
1 steve Male Pre Explain how you would... "Response"
2 jenna Female Pre Which of the following... "Response 1"
2 jenna Female Pre Which of the following... "Response 2"
2 jenna Female Pre Please indicate your race. "White"
2 jenna Female Pre Explain how you would... "Response"
1 jack Male Pre Which of the following... "Response 1"
1 jack Male Pre Which of the following... "Response 2"
1 jack Male Pre Which of the following... "Response 3"
1 jack Male Pre Please indicate your race. "Black"
1 jack Male Pre Explain how you would... "Response"
3 billy Male Pre Which of the following... "Response 1"
3 billy Male Pre Which of the following... "Response 2"
3 billy Male Pre Please indicate your race. "Other"
3 billy Male Pre Explain how you would... "Response"
当我对数据进行格式化时，数据更有用，因此每个单独的问题在数据框中都有自己的列，因此，通过协作，我们编写了以下代码，将每个独特的问题转换为自己的列：
x = data
require(stringr)
temp = which(x$Db == "Db")
x = x[-temp,]
questions = unique(x$Question)
concdMeta = apply(x[,c('Db', 'username', 'gender', 'interviewForm')],
MARGIN = 1,
FUN = paste,
collapse = " & ")
students = unique(concdMeta)
out = matrix(nrow = length(students),
ncol = 4 + length(questions))
row.names(out) = students
colnames(out) = c(colnames(x)[1:4], questions)
out = data.frame(out)
for(i in 1:nrow(out)) {
Z = str_split(row.names(out)[i], ' & ')[[1]]
for(j in 1:4) {
out[i, j] = Z[j]
}
}
for (i in 1:nrow(x))
{
db = x$Db[i]
un = x$username[i]
g = x$gender[i]
iF = x$interviewForm[i]
q = x$Question[i]
a = x$Answer[i]
this.meta = paste(x[i, 1:4], collapse = " & ")
matching.row = row.names(out) == this.meta
out[matching.row, 4 + which(questions == q)] = a
}
out[is.na(out)] = "NA"
上面的代码非常有效，除了我们遇到的一个问题：它列出了对问题的最终回答，无论最后一个回答是针对该学生的。因此，对于，比如史蒂夫上面，它只会列出“响应3”或詹纳的“响应2”等。如此：
>data.reformatted
Database Username Gender InterviewType Which of the following... Please indicate... Explain how...
1 steve Male Pre "Response 3" "White" "Response"
2 jenna Female Pre "Response 2" "White" "Response"
1 jack Male Pre "Response 3" "Black" "Response"
3 billy Male Pre "Response 2" "Other" "Response"
我们可以对上面的代码进行添加以使其将所有响应连接到单个单元格中吗？意味着最终数据集看起来像这样：
Database Username Gender InterviewType Which of the following... Please indicate... Explain how...
1 steve Male Pre "Response 1, Response 2, Response 3" "White" "Response"
2 jenna Female Pre "Response 1, Response 2" "White" "Response"
1 jack Male Pre "Response 1, Response 2, Response 3" "Black" "Response"
3 billy Male Pre "Response 1, Response 2" "Other" "Response"
答案 0 :(得分：1)
除了上面的答案，更自动化的版本将是：
library(dplyr)
library(tidyr)
data$rownum <- 1:nrow(data) # add row numbers as a column
# needed for the spread function to work
# (complains about duplicates otherwise)
questions <- as.character(unique(data$Question))
qNames <- paste0("q", 1:length(questions))
data <- data.frame(lapply(data, as.character), stringsAsFactors = FALSE)
# change questions names to question keys
for (q in 1:length(questions)){
data[data$Question == questions[q], "Question"] <- qNames[q]
}
data.wide <- data %>%
spread(Question, Answer)
#colnames(data.wide)[6:8] <- c("Explain", "Indicate", "Which") # change column names
data.wide <- data.frame(lapply(data.wide, as.character), stringsAsFactors = FALSE)
data.wide[is.na(data.wide)] <- ""
head(data.wide)
# Group by necessary columns and merge rows using Reduce and paste
dummy <- data.wide %>%
dplyr::group_by(Database, Username, Gender, InterviewType)
# Create command as a string
str_start <- "dummy2 <- dummy %>% dplyr::summarize("
str_end <- paste0(qNames[length(qNames)]," = Reduce(function(...) paste(...),", qNames[length(qNames)],"))")
str_middle <- ""
for (q in 1:(length(qNames) - 1 )){
str_middle <- paste(str_middle, qNames[q] ," = Reduce(function(...) paste(...),", qNames[q], "),")
}
str_final <- paste0(str_start, str_middle, str_end)
# Execute Command
eval(parse(text = str_final))
# Change columns names
for (q in 1:length(questions)){
colnames(dummy2)[colnames(dummy2) == qNames[q]] <- as.character(questions[q])
}
答案 1 :(得分：0)
我已经把一些似乎做你想做的代码放在一起了。
您需要安装dplyr才能运行此功能。
data$rownum <- 1:nrow(data) # add row numbers as a column
# needed for the spread function to work
# (complains about duplicates otherwise)
data.wide <- data %>%
spread(Question, Answer)
colnames(data.wide)[6:8] <- c("Explain", "Indicate", "Which") # change column names
data.wide[is.na(data.wide)] <- ""
head(data.wide)
Database Username Gender InterviewType rownum Explain Indicate Which
1 1 jack Male Pre 10 "Response 1"
2 1 jack Male Pre 11 "Response 2"
3 1 jack Male Pre 12 "Response 3"
4 1 jack Male Pre 13 "Black"
5 1 jack Male Pre 14 "Response"
6 1 steve Male Pre 1 "Response 1
# Group by necessary columns and merge rows using Reduce and paste
result <- data.wide %>%
dplyr::group_by(Database, Username, Gender, InterviewType) %>%
dplyr::summarize(
Which = Reduce(function(...) paste(...), Which),
Indicate = Reduce(function(...) paste(...), Indicate),
Explain = Reduce(function(...) paste(...), Explain))
head(as.data.frame(result))
Database Username Gender InterviewType Which Indicate Explain
1 1 jack Male Pre "Response 1" "Response 2" "Response 3" "Black" "Response"
2 1 steve Male Pre "Response 1" "Response 2" "Response 3" "White" "Response"
3 2 jenna Female Pre "Response 1" "Response 2" "White" "Response"
4 3 billy Male Pre "Response 1" "Response 2" "Other" "Response"
希望这会有所帮助。