Data Visualization with R: Homework 2

This is my product for homework 2 in this course.

rm(list = ls())

# Loading packages
library(dplyr)
library(readxl)
library(ggplot2)
library(janitor)
library(stringr)
library(stringi)

# Reading data
readxl::read_xlsx("C:/Users/Phong Vu/Dropbox/PC/Downloads/r_cased/R15_day2/_data_pci_file_general_file_1588582407-Indicator_PCI2018_VN.xlsx") -> pci_2018

# Cleaning and Wrangling data
pci_2018  %>% 
  clean_names() %>% 
  select(tinh_thanh_pho, xep_hang, diem_so_pci_2018) %>%
  slice(-64,-65, -66, -67, -68, -69) %>%
  rename(province = tinh_thanh_pho,
         rank = xep_hang,
         pci = diem_so_pci_2018) -> pci_2018

# Visualization
pci_2018 %>% 
  mutate(province_rank = case_when(rank < 10 ~ str_c(province, " ", "0", rank),
                                   TRUE ~ str_c(province, " ", rank))) -> pci_2018

pci_2018$pci <- round(pci_2018$pci, 2)

pci_2018 %>% 
  mutate(class = case_when(pci >= 70 ~ "Excellent",
                           pci >= 65.40 & pci < 70 ~ "Good",
                           pci >= 62.40 & pci < 65.40 ~ "Fair",
                           pci >= 60.02 & pci < 62.40 ~ "Mediore",
                           pci < 60.02 ~ "Poor")) -> pci_2018

pci_2018 %>%
  arrange(pci) %>%
  mutate(province_rank_fac = factor(province_rank, levels = province_rank)) -> pci_2018

pci_2018 %>% 
  mutate(class_color = case_when(class == "Excellent" ~ "#470150", 
                                 class == "Good" ~ "#3c558d",
                                 class == "Fair" ~ "#238f89",
                                 class == "Mediore" ~ "#5bc866", 
                                 class == "Poor" ~ "#f3e933")) %>% 
  pull(class_color) -> color_on_class

pci_2018 %>% 
  ggplot(aes(x = pci, y = province_rank_fac)) +
  geom_col(width = 0.8, fill = color_on_class) +
  geom_text(aes(label = pci), 
            size = 1.55, 
            position = position_dodge(0.5), 
            hjust = -0.2) +
  theme(axis.ticks = element_blank(),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        axis.title = element_blank(),
        axis.text.y = element_text(size = 4, color = "black"),
        axis.text.x = element_blank(),
        legend.position = c(0.95, 0.95),
        legend.box.just = "right",
        panel.border = element_blank(),
        panel.background = element_blank(),
        plot.title = element_text(face = "bold"),
        plot.subtitle = element_text(color = "gray50"),
        plot.caption = element_text(color = "gray70", face = "italic")) +
  scale_x_continuous(expand = c(0, 0), 
                     breaks = seq(0, 80), 
                     limits = c(0, 80)) +
  labs(title = "Vietnam PCI Index 2018",
       subtitle = "R Used for Data Visualization",
       caption = "Data source: https://pcivietnam.vn/du-lieu-pci")