MATH 216 Homework 2

Admistrative:
Question 1:
Income
Job
Orientation
Question 2:

## Warning: package 'ggplot2' was built under R version 3.2.4

Admistrative:

Please indicate

Who you collaborated with:
Roughly how much time you spent on this HW so far:
The URL of the RPubs published URL here.
What gave you the most trouble:
Any comments you have:

Question 1:

Perform an Exploratory Data Analysis (EDA) on the profiles data set, specifically on the relationship between gender and

income
job
One more categorical variable of your choice

all keeping in mind in HW-3, you will be fitting a logistic regression to predict a user’s gender based on these variables.

Income

The following will analyze the relationship between gender and income.

profiles <- profiles %>%
  mutate(is_female = ifelse(sex=="f", 1, 0))

profiles <- profiles %>%
  mutate(income_bracket = ifelse(income == -1, "unreported", 
                                 ifelse(income <= 50000, "under_50k",
                                        ifelse(income <= 150000, "50k_to_150k",
                                               ifelse(income <= 10000000, 
                                                      "150k_to_10mil", "over_10mil")))))

compare_incomes <- profiles %>%
  group_by(income_bracket, sex) %>%
  summarise(count=n()) %>%
  group_by(income_bracket) %>%
  mutate(prop=count/sum(count))

compare_incomes %>% knitr::kable(digits = 4)

income_bracket	sex	count	prop
150k_to_10mil	f	133	0.1852
150k_to_10mil	m	585	0.8148
50k_to_150k	f	979	0.2037
50k_to_150k	m	3827	0.7963
under_50k	f	2001	0.3346
under_50k	m	3979	0.6654
unreported	f	21004	0.4336
unreported	m	27438	0.5664

ggplot(compare_incomes, aes(x = income_bracket, y = prop, fill = sex)) +
  geom_bar(stat="identity", position = "dodge") +
  ggtitle("Self reported incomes of SF OkCupid users")

# income_hist<- profiles %>% 
#   filter(income != -1) %>%
#   ggplot(aes(x = income)) +
#   geom_histogram(bins = 30) +
#   facet_wrap(~sex) +
#   ggtitle("Self reported income by gender in SF OkCupid users")
# income_hist

profiles %>% filter(income!= -1) %>% summarise(mean(income))

## Source: local data frame [1 x 1]
## 
##   mean(income)
##          (dbl)
## 1       104395

no_reported_income <-  profiles %>% 
  filter(income == -1) %>%
  summarise(mean(sex == 'f'), mean(sex == 'm'))

no_reported_income %>% knitr::kable(digits = 4, 
                                    caption = "Gender breakdown for missing income values",
                                    col.names = c("proportion female", "proportion male"))

Gender breakdown for missing income values
proportion female	proportion male
0.4336	0.5664

mean(profiles$income == -1)

## [1] 0.808094

not_reporting_income <- profiles %>%
  group_by(sex) %>%
  summarise("Proportion not reporting income" = mean(income == -1))
not_reporting_income %>% knitr::kable(digits = 2)

sex	Proportion not reporting income
f	0.87
m	0.77

80.8% of users did note report an income. 87.09% of women didn’t report and 76.58% of men didn’t report. The “unreported” category is the only category in which females make up more that 40% of the category, even though 40% of users in the overall data are female.

Job

The following will analyze the relationship between gender and job.

n_distinct(profiles$job)

## [1] 22

ggplot(profiles, aes(x=job)) +geom_bar(stat="count")

test <- profiles %>% 
  group_by(job) %>%
  summarise(count = n()) %>%
  ungroup()%>%
  mutate(job_new =ifelse(count < 600, "misc", 
                         ifelse(job == "", "unreported", job)))

profiles <- left_join( profiles, test, by = "job") %>% select(-count)

jobs_hist<- profiles %>% 
  group_by(job_new, sex) %>% 
  summarise(count=n()) %>% 
  group_by(job_new) %>% 
  mutate(prop=count/sum(count))

ggplot(jobs_hist, aes(x = job_new, y=prop, fill= sex)) +
  geom_bar(stat="identity", position = "dodge") +
  ggtitle("Jobs of SF OkCupid users") +
  theme(axis.text.x=element_text(angle=90))

profiles %>% group_by(job_new) %>% 
  summarise(percent_female = mean(is_female)) %>% 
  knitr::kable(digits = 4)

job_new	percent_female
artistic / musical / writer	0.4240
banking / financial / real estate	0.3486
clerical / administrative	0.7106
computer / hardware / software	0.1361
construction / craftsmanship	0.0793
education / academia	0.6134
entertainment / media	0.3027
executive / management	0.3182
hospitality / travel	0.4179
law / legal services	0.4634
medicine / health	0.5943
misc	0.3074
other	0.4802
political / government	0.4379
sales / marketing / biz dev	0.4363
science / tech / engineering	0.2096
student	0.4406
unreported	0.4460

not_reporting_job <- profiles %>%
  group_by(sex) %>%
  summarise("Proportion not reporting job" = mean(job == "")) %>%
  knitr::kable(digits = 2)
not_reporting_job

sex	Proportion not reporting job
f	0.15
m	0.13

sum(profiles$job == "")/sum(!is.na(profiles$job)) #overall not reporting job

## [1] 0.1367564

job_reporting_other <- profiles %>%
  group_by(sex) %>%
  summarise("Percent reporting job as other" = mean(job == "other")) %>%
  knitr::kable(digits = 2)
job_reporting_other

sex	Percent reporting job as other
f	0.15
m	0.11

sum(profiles$job == "other")/sum(!is.na(profiles$job))

## [1] 0.1265973

computer_dudes <- profiles %>%
  group_by(sex) %>% 
  summarise("percent in computers" = mean(job == "computer / hardware / software"))

computer_dudes %>% knitr::kable(digits=4)

sex	percent in computers
f	0.0266
m	0.1135

tech_dudes <- profiles %>%
  group_by(sex) %>% 
  summarise("percent in science/tech/engineering" = 
              mean(job == "science / tech / engineering"))

tech_dudes %>% knitr::kable(digits=4)

sex	percent in science/tech/engineering
f	0.0421
m	0.1070

Overall, 13.7% of users did not report a job. 15.16% of females and 12.68% of males didn’t report. 12.6% of users fall under the category other; 15% of females and 11% of males. Also, 11.3% of males are in computers/hardware/software, while only 2.6% of women are. Similar numbers for the STEM category. This mean that 21% of men are in one of these 2 categories but only 6% of women are.

Also, I lumped all jobs that constituted less than 1% of users (count < 600) into one category “misc.”

Orientation

n_distinct(profiles$orientation)

## [1] 3

sexual_orientation_bar <- profiles %>%
  group_by(orientation, sex) %>%
  summarise(count = n()) %>%
  group_by(orientation) %>%
  mutate(prop = count/sum(count))

ggplot(sexual_orientation_bar, aes(x = orientation, y = prop, fill = sex)) +
  geom_bar(position = "dodge", stat="identity") +
  ggtitle("Sexual Orientation in SF OKCupid Users")

orientation <-  profiles %>% 
  group_by(orientation) %>%
  summarise("female" = mean(sex == 'f'),
            "male" = mean(sex == 'm')) %>%
  knitr::kable(digits=4)
orientation

orientation	female	male
bisexual	0.7214	0.2786
gay	0.2849	0.7151
straight	0.3979	0.6021

Although only 40% of users are female, 72% of users who are bisexual are women.

Question 2:

In the file HW-2_Shiny_App.Rmd, build the Shiny App discussed in Lec09 on Monday 10/3: Using the movies data set in the ggplot2movies data set, make a Shiny app that

Plots budget on the x-axis and rating on the y-axis
Instead of having a radio button to select the genre of movie (Action, Animation, Comedy, etc), have a radio button that allows you to toggle between comedies and non-comedies. This app should be simpler.