Overview
In this take-home exercise, appropriate static statistical graphics methods are used to reveal the demographic of the city of Engagement, Ohio USA. The data was processed by using appropriate tidyverse family of packages and the statistical graphics were prepared using ggplot2 and its extensions.
Getting Started
Before we get started, it is important for us to ensure that the required R packages have been installed. If yes, we will load the R packages. If they have yet to be installed, we will install the R packages and load them onto R environment.
The chunk code below will do the trick.
packages = c('tidyverse', 'ggdist', 'gghalves', 'ggridges','knitr', 'ggpubr')
for(p in packages){
if(!require(p, character.only = T)){
install.packages(p)
}
library(p, character.only = T)
}
Importing Data
The code chunk below imports Participants.csv from the data
folder into R by using the read_csv()
function of readr
and saves it as a tibble data frame called participants.
participants <- read_csv("data/Participants.csv")
Data
The code chunk below shows a preview of the data table using the kable()
function from knitr.
kable(participants[1:5, ], caption = 'Participants data table', align = "l")
| participantId | householdSize | haveKids | age | educationLevel | interestGroup | joviality |
|---|---|---|---|---|---|---|
| 0 | 3 | TRUE | 36 | HighSchoolOrCollege | H | 0.0016267 |
| 1 | 3 | TRUE | 25 | HighSchoolOrCollege | B | 0.3280865 |
| 2 | 3 | TRUE | 35 | HighSchoolOrCollege | A | 0.3934696 |
| 3 | 3 | TRUE | 21 | HighSchoolOrCollege | I | 0.1380634 |
| 4 | 3 | TRUE | 43 | Bachelors | H | 0.8573967 |
The data table has 1,011 rows and 7 columns.
Analysis
The scope of this analysis is to visualise the relationship between joviality and the rest of the participant’s characteristics. From this we might discover trends that can teach us about what makes a person jovial. The following sections show the various methods used to visualise this relationship and the insights we can learn from them.
Distribution of Joviality
The code chunk below plots a density plot for joviality by using the
geom_density()
of ggplot2.
ggplot(data=participants,
aes(x = joviality)) +
geom_density() +
ggtitle("Figure 1: What is the distribution of Joviality?")
The joviality for each person is measured between a scale of 0 to 1, from the density plot we can see that joviality among the participants is somewhat uniformly distributed.
Joviality vs Having Kids
The code chunk below plots two kernel density lines for joviality by using colour or fill arguments of aes(). One line is for participants with kids while the other is for those without.
ggplot(data=participants,
aes(x = joviality,
colour = haveKids)) +
geom_density() +
ggtitle("Figure 2: Do kids make people jovial?")
While there is large overlap between the distribution of joviality for both groups, we can see that people with kids tend to be more jovial than those without. However, this trend reverses when looking at participants at the high end of joviality (above 0.9).
Joviality vs Household Size
The code chunk below combines a violin plot and a boxplot to show the
distribution of joviality by householdSize. This is
done using geom_violin()
and geom_boxplot().
ggplot(data=participants,
aes(y = joviality,
x= as.character(householdSize))) +
xlab("house hold size") +
geom_violin(fill="light blue") +
geom_boxplot(alpha=0.5) +
stat_summary(geom = "point",
fun.y="mean",
colour ="red",
size=2) +
ggtitle("Figure 3: Is 'the more the merrier' true?")
The figure above shows that having a householdSize > 1 does result in a higher joviality. However, what is does not show is that larger families are more jovial. While mean joviality for householdSize of 2 and 3 are similar, the median joviality for a householdSize of 2 is slightly higher. The violin plot also shows a large bump at the high end of joviality for participants with householdSize of 1. This could mean that there exist a group of extremely jovial individuals that prefer living by themselves.
Joviality vs Age
In the code chunk below, geom_point()
is used to create the scatterplot while geom_smooth()
is used to plot a best fit line. The R2 value is calculated
using the stat_regline_equation()
function from ggpubr
ggplot(data=participants,
aes(x= age,
y=joviality)) +
geom_point() +
geom_smooth(method="lm",
size=0.5) +
stat_regline_equation(label.y = 1.1, aes(label = ..rr.label..)) +
ggtitle("Figure 4: Does age come with joviality?")
The figure above show a wide spectrum of joviality across all ages. While the best fit line shows a negative correlation between age and joviality, this relationship is insignificant with an R2 value of only 0.0047
Joviality vs Education Level
The code chunk below plots boxplots of joviality against
educationLevel. Points representing the mean values are added
using the stat_summary()
function. lastly, the boxplots are ordered ascendingly in terms of the
level of education using the fct_relevel()
function.
participants %>%
mutate(educationLevel = fct_relevel(educationLevel,
"Low", "HighSchoolOrCollege", "Bachelors",
"Graduate")) %>%
ggplot(aes(y = joviality,
x= educationLevel)) +
geom_boxplot(notch=TRUE) +
stat_summary(geom = "point",
fun.y="mean",
colour ="red",
size=2) +
stat_summary(geom = "text",
aes(label=paste("mean = ",round(..y..,2))),
fun.y="mean",
colour="red",
size=4,
vjust = -2) +
ggtitle("Figure 5: Does education affect joviality?")
From the figure above, we can see that participants with low/highschool/college/bachelors education have similar mean and median joviality (as shown by the overlapping notches) while graduates have a higher mean and median joviality than the rest.
Joviality vs Interest Group
The code chunk below plots boxplots of joviality against
interestGroup. As the interestGroup alphabet is
arbitrary, the plot is instead ordered according to each groups mean
joviality using the reorder()
function.
ggplot(data=participants, aes(y = joviality,
x= reorder(interestGroup, joviality, FUN = mean))) +
xlab("interestGroup") +
geom_boxplot() +
stat_summary(geom = "point",
fun.y="mean",
colour ="red",
size=2) +
stat_summary(geom = "text",
aes(label=paste("mean=",round(..y..,2))),
fun.y="mean",
colour="red",
size=2,
vjust = -2) +
ggtitle("Figure 6: Which interest group is the most jovial?")
From the figure above, we can see that the top 3 interestGroup in terms of mean joviality are groups E, C and G while the bottom 3 are groups H, A and D.
Conclusion
The visual techniques above show that in general, joviality is higher with participants with kids, with a householdSize >1, who are graduates, and who belonging to interestGroup E, C and G. Lastly, we see that age has little effect on the joviality of the individuals.