Take-Home-Exercise_02

For this Take-Home Exercise 2, I would like to introduce how we can make the visualization of take-home exercise 1 better in terms of clarity and aesthetics.

Problem Statement

For a recap, exercise 1’s task was to create a trellis chart of age & gender pyramids in 9 selected planning areas.

Peer_Critic

How can we improve visualization?

Although this visualization has managed to plot pyramids by different planning area, there are still some rooms to improve to make it look better!

Clarity

From the perspective of the audience who first read this chart, they may not be familiar with the terms used in the chart. Therefore, it would be desirable to spell out acronyms, such as “Planning Area” instead of “PA”. In addition, title can also be improved to deliver the original intention of the task. For example, simple and clear title such as “Singapore’s Age and Sex Pattern by Planning Area” could be good enough. Last but not least, adding footnotes or descriptions to make the chart more understandable could be another way to improve the clarity of this visualization.

Aesthetics

There are mainly two issues with this visualization: First, it is hard to tell female and male bars from each pyramid. Second, repeated labels make the visualization less readable. In this article, I will suggest an improved visualization by using ggplot and tidyverse.

Importing the data

First, start from loading tidyverse and importing dataset.

Show the code 
pacman::p_load(tidyverse)

In this task, I used Singapore Residents by Planning Area / Subzone, Age Group, Sex and Type of Dwelling, June 2022 published by Department of Statistics, Singapore.

I removed null data and wrong category from the csv file, and rebinded male and female dataset.

Show the code 
# prepare cleaned data
male_pop <- read_csv("data/Male_Pop_June_2022.csv")
Rows: 24003 Columns: 7
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (5): Sex, PA, Subzone, AG, Floor Area
dbl (1): Time
num (1): Population

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
Show the code 
female_pop <- read_csv("data/Female_Pop_June_2022.csv")
Rows: 24228 Columns: 7
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (5): Sex, PA, Subzone, AG, Floor Area
dbl (1): Time
num (1): Population

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
Show the code 
# removed null data, wrong category
# bind and inspect data

total_pop <- rbind(male_pop, female_pop)

Data Cleaning Result

Then, need to aggregate and leave necessary columns only for easier analysis.

Show the code 
freq_pop <- total_pop %>%
  group_by(`Sex`, `AG`, `PA`) %>%
  summarise('Count'= sum(`Population`)) %>%
  ungroup()
`summarise()` has grouped output by 'Sex', 'AG'. You can override using the
`.groups` argument.

Frequency Aggregation

In this practice, I chose Ang Mo Kio, Bedok, Bukit Panjang, Clementi, Choa Chu Kang, Hougang, Jurong East, Serangoon, and Tampines. Here are steps to improve visualization, with Ang Mo Kio example.

Step-by-Step Plotting

Ang Mo Kio

Filter the dataset by PA == “Ang Mo Kio”, then see how the plot looks like with female dataset.

Show the code 
# get AMK first

amk_pop <- freq_pop %>%
  filter(PA == "Ang Mo Kio")

amk_pop_female <- freq_pop %>%
  filter(Sex == "Female", PA == "Ang Mo Kio")

ggplot(amk_pop_female,
       aes(x = Count,
           y = AG,
           fill = PA)) + 
  geom_col()

Let’s see how to plot male data. You may use convert the x axis in negative value to switch the axis direction.

Show the code 
amk_pop_male <- freq_pop %>%
  filter(Sex == "Male", PA == "Ang Mo Kio")

ggplot(amk_pop_male,
       aes(x = -Count,
           y = AG,
           fill = PA)) + 
  geom_col()

Now, let’s put them together and see how age & sex pyramid looks like.

Show the code 
amk_pyramid <- amk_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

amk_plot <- 
  ggplot(amk_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex)) +
    geom_col() 

amk_plot # to get the final pyramid

As we will need to put each chart together, let’s remove overlapping labels and legend.

Show the code 
amk_plot <- 
  ggplot(amk_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Ang Mo Kio")

amk_plot

Similarly, you can create 8 other plots using the same method. In this practice, I chose Bedok, Bukit Panjang, Clementi, Choa Chu Kang, Hougang, Jurong East, Serangoon, and Tampines.

Bedok

Show the code 
# Bedok
bedok_pop <- freq_pop %>%
  filter(PA == "Bedok")

bedok_pyramid <- bedok_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

bedok_plot <- 
  ggplot(bedok_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Bedok")

Bukit Panjang

Show the code 
# Bukit Panjang
bk_pj_pop <- freq_pop %>%
  filter(PA == "Bukit Panjang")

bk_pj_pyramid <- bk_pj_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

bk_pj_plot <- 
  ggplot(bk_pj_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Bukit Panjang")

Clementi

Show the code 
# Clementi
clementi_pop <- freq_pop %>%
  filter(PA == "Clementi")

clementi_pyramid <- clementi_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

clementi_plot <- 
  ggplot(clementi_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Clementi")

Choa Chu Kang

Show the code 
# Choa Chu Kang
cck_pop <- freq_pop %>%
  filter(PA == "Choa Chu Kang")

cck_pyramid <- cck_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

cck_plot <- 
  ggplot(cck_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Choa Chu Kang")

Hougang

Show the code 
# Hougang
hougang_pop <- freq_pop %>%
  filter(PA == "Hougang")

hougang_pyramid <- hougang_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

hougang_plot <- 
  ggplot(hougang_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Hougang")

Jurong East

Show the code 
# Jurong East
jr_est_pop <- freq_pop %>%
  filter(PA == "Jurong East")

jr_est_pyramid <- jr_est_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

jr_est_plot <- 
  ggplot(jr_est_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Jurong East")

Serangoon

Show the code 
# Serangoon
srgoon_pop <- freq_pop %>%
  filter(PA == "Serangoon")

srgoon_pyramid <- srgoon_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

srgoon_plot <- 
  ggplot(srgoon_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Serangoon")

Tampines

Show the code 
# Tampines

tamp_pop <- freq_pop %>%
  filter(PA == "Tampines")

tamp_pyramid <- tamp_pop %>%
  mutate(
    Count = case_when(
      Sex == "Male" ~ -Count,
      TRUE ~ Count
    ))

tamp_plot <- 
  ggplot(tamp_pyramid,
         aes(x = Count,
             y = AG,
             fill = Sex), show.legend=FALSE) +
    geom_col() +
    theme(axis.title.x = element_blank(),
          axis.title.y = element_blank(),
          legend.position = "none") + 
    ggtitle("Tampines")

Trellis Chart

Now we have 9 separate pyramids, let’s put them in one view using patchwork.

You may use following code to download patchwork package.

devtools::install_github(“thomasp85/patchwork”)

Show the code 
library(ggplot2)
library(patchwork)

Once you have patchwork ready, you can put 9 pyramids together, and add title!

Show the code 
patchwork <- (amk_plot | bedok_plot | bk_pj_plot)/
  (clementi_plot | cck_plot | hougang_plot)/
  (jr_est_plot | srgoon_plot | tamp_plot)

patchwork + plot_annotation(
  title = 'Singapore Age and Sex Pattern by Planning Area')

In this exercise, the initial visualization has been improved in terms of clarity and aesthetics by:

  • Adding appropriate title and planning area

  • Separating female and male with different colors for easier comparison

  • Removing axis title for cleaner view

  • Putting 9 pyramid plots in 1 view