R：多用户响应数据帧上的直方图和密度

Question

Description of Data

Data reflect how users rated a book on an online book recommendation site while answering a question which has four answers. Users were allowed to choose more than one answer.

Goal is to obtain distribution plots by gender where X axis as answer (X1,X2..) and Y axis as the count of books along with density overlay. It would be great for both male and female to be overlay one another.

book_id  user_id  rate  X1   X2   X3    X4  Gender  genre
40         001     4.5    0    1    0    0  male    fiction
48         001     3.5    1    0    0    1  male    fiction
54         001     4.0    1    0    0    0  male    fiction
79         001     2.5    1    0    1    0  male    non-fiction
80         001     4.5    0    0    1    0  male    non-fiction
95         001     5.0    1    0    1    0  male    non-fiction
95         002     3.0    0    0    0    1  Female  non-fiction
99         002     4.5    0    0    1    0  Female  non-fiction
02         002     0.5    0    0    0    0  Female  non-fiction
05         002     4.5    1    0    1    0  Female  non-fiction
54         002     4.0    0    1    0    0  Female  fiction
79         002     2.5    1    0    1    0  Female  non-fiction
80         002     4.5    0    0    1    0  Female  non-fiction
07         002     4.5    1    0    1    0  Female  fiction
07         003     5.0    1    0    1    0  Female  fiction
09         003     4.0    0    0    1    0  Female  auto-bio
54         003     4.0    1    0    0    0  Female  fiction
79         003     2.5    1    0    1    0  Female  non-fiction
80         003     4.5    0    0    1    0  Female  non-fction
17         004     3.5    1    0    0    0  male    auto-bio
21         004     5.0    1    0    1    0  male    auto-bio
21         005     5.0    0    1    1    0  male    auto-bio
17         005     0.5    0    0    0    1  male    auto-bio
20         005     5.0    0    0    1    0  male    fiction
20         006     1.5    0    0    0    1  male    fiction
21         006     5.0    0    0    1    0  male    auto-bio
21         007     2.0    1    0    0    0  male    auto-bio
21         008     4.5    1    0    1    0  Female  auto-bio
20         008     4.5    1    0    1    0  Female  fiction
07         008     4.5    1    0    1    0  Female  fiction
22         009     5.0    0    0    1    0  male    fiction
54         009     4.0    1    0    0    0  male    fiction
79         009     2.5    1    0    1    0  male    non-fiction
80         010     4.5    1    0    1    0  male    non-fiction
22         010     4.5    0    1    1    0  male    fiction
22         011     0.5    0    0    1    0  Female  fiction
28         011     3.5    1    0    0    0  Female  auto-bio

Two users can rate the same book and answer the question in the same way or different way. This creates two records per each book. With that in mind, If group by Gender and sum each column down would give gender level distribution to start with.

df %>% group_by(Gender) %>% summarize(x1 = sum(X1), x2 = sum(X2), x3=sum(X3),x4 =sum(X4))

  Gender    x1    x2    x3    x4
  <fct>  <int> <int> <int> <int>
1 Female    10     1    13     1
2 male      10     3    11     3

In addition to getting the plot: I also have the following question: Also just to confirm this is not the unique number of books female answer x1 since the same book can be answered by multiple users. Instead, it would be number of female choose a specific answer?

Answer 1

A similar but different approach

library(data.table)
library(ggplot2)
dt <- setDT(dt)

plottest <- melt(dt,measure.vars = patterns("^X"),variable.name = "question", value.name = "answer")

ggplot(data = plottest,aes(factor(book_id),answer))+
  geom_col(aes(fill = as.factor(question), color = as.factor(question) ))+
  facet_wrap(~Gender)+
  labs(title =  "",
       y = "N",
       x = "books",
       color = "Question",
       fill = "Question")

Answer 2

I am not sure I understand correctly but is the following code what you want?

library(dplyr)
library(ggplot2)

df2 <- df %>% 
  group_by(Gender) %>% 
  summarize(x1 = sum(X1), x2 = sum(X2), x3=sum(X3),x4 =sum(X4)) %>%
  melt(id.vars = "Gender")


ggplot(df2, aes(variable, value, color = Gender, fill = Gender)) +
  geom_bar(stat = "identity", position = "dodge")

After seeing the answer by @denis I adapted his code to do more or less the same but with position = "dodge" .

df3 <- df %>% 
  group_by(Gender, book_id) %>% 
  summarize(x1 = sum(X1), x2 = sum(X2), x3=sum(X3),x4 =sum(X4)) %>%
  melt(id.vars = c("Gender", "book_id"))

ggplot(df3, aes(as.factor(book_id), value, color = variable, fill = variable)) +
  geom_bar(stat = "identity", position = "dodge") +
  facet_wrap(~ Gender)

As for the second question, you can use aggregate to get the answers to each question by Gender .

aggregate(. ~ Gender, df[4:8], sum)
#  Gender X1 X2 X3 X4
#1 Female 10  1 13  1
#2   male 10  3 11  3