data.R

##Marine Fujisawa

## Sample Project: Chosen because it includes data cleaning, web scraping, data visualizations, and a code that produces a ShinyApp to allow the user to interact with the data.  

# The first step I take is the collection and cleaning of data. 
# After obtaining the data, it was then cleaned to ensure consistency across the different datasets.
# Next, I created a scatterplot to visualize some relationships within the data. I also create a choropleth map to be used for the ShinyApp. 

# Setting work directory
setwd("/Users/marinefujisawa1/Documents/GitHub/Coding-Sample")
#load libraries
library(tidyverse)
library(dplyr)

# Import first dataset: GDP of countries
gdp<-read.csv("data/GDP_countries.csv")
#View(gdp)

# Clean data
# First, remove last 5 rows as they are blank
gdp<-head(gdp,-5)

#We want to know economic recovery from the COVID crisis, so only going to keep 2019-2023 data
gdp <- gdp[, -c(5:11)]

#Next, replace ".." values with NA
gdp <- gdp %>%
  mutate_all(~ na_if(., ".."))

#Change column names of year so it's easier to use
years<-c("2019", "2020", "2021", "2022", "2023")
colnames(gdp)[(ncol(gdp)-4):ncol(gdp)] <- years

#Convert gdp values to numeric
gdp <- gdp %>%
  mutate(across((ncol(.)-4):ncol(.), as.numeric))

## Add some new metrics to measure how well each country recovered from COVID shocks
#Recovery percentage- Using 2019 GDP as the baseline, I use the formula: 
#[(GDPyear - GDP2020)/(GDP2019 - GDP 2020)]*100%
gdp<-gdp%>%
  mutate(recovery2023 = ((`2023`-`2020`)/(`2019`-`2020`))*100)

#Also want to measure how much of a shock they experienced
#(GDP2019-GDP2020)/GDP2019*100
gdp<-gdp%>%
  mutate(decline = ((`2019`-`2020`)/`2019`)*100)


# Import second dataset: Export diversification index
export<-read.csv("data/Export_diversification_index.csv")
#View(export)

#Reducing number of columns so it's easier to see- we only need 2019 diversification data
export<-export[, -c(2:57)]

#Removing first row since it contains no data
export <- export[-1, ]

#Adding new column with country code so it can merge better with gdp data
#install.packages("countrycode")
library(countrycode)
export$country_iso <- countrycode(export$Economy_Label, "country.name", "iso3c")

#Again, cleaning up some columns since the data is not needed
export<-export[,-c(4:17)]
#Missed one column so dropping here
export <- export %>% select(-X2018_Diversification_Index_MissingValue)

#Now, will use webscraping to get another economic diversification index
library(rvest)
url<-"https://economicdiversification.com/the-index/#"
index_website<-read_html(url)
econ_diverse <- html_table(index_website, fill = TRUE)

#Loading the dataset, cleaning
econ_diverse<-(econ_diverse[[1]])

#adding the country code
econ_diverse$country_iso <- countrycode(econ_diverse$Countries, "country.name", "iso3c")

#Finally, merging all data into one dataset
gdp <- gdp %>%
  rename(country_iso = Country.Code)
initial_merge <- merge(gdp, export, by = "country_iso")

final_data<-merge(initial_merge, econ_diverse, by = "country_iso")
View(final_data)

write.csv(final_data, "final_data.csv", row.names = FALSE)

## Next, I created some scatterplots. Here, I am showing one of them (so the file doesn't get too long). 

# First plot: scatterplot, using export diversification
# I wanted to create a plot without outliers.
# # Define outliers as (values outside 1.5*IQR from Q1 and Q3)
Q1 <- quantile(final_data$recovery2023, 0.25, na.rm=TRUE)
Q3 <- quantile(final_data$recovery2023, 0.75, na.rm=TRUE)
IQR_value <- Q3 - Q1

final_data_no_outliers <- final_data %>%
  filter(recovery2023 >= (Q1 - 2 * IQR_value) & recovery2023 <= (Q3 + 2 * IQR_value))

# Plot without outliers
plot1<-ggplot(final_data_no_outliers, aes(x = X2019_Diversification_Index_Value, y = decline)) +
  geom_point(color="olivedrab") +
  geom_smooth(method = "lm", color = "tan4")+
  labs(title = "Export Diversificaiton vs GDP Decline Percentage Post COVID", x = "Export Diversification", y = "GDP Decline Percentage")+
  theme_minimal()+
  theme(plot.title = element_text(size = 13, face = "bold", hjust = 0.5))

plot1

ggsave(plot1, 
       filename = "images/plot1.png",
       device = "png")

# Next, I created a choropleth map to be used later.
#install.packages("sf")
library(sf)
#install.packages("spData")
library(spData)

world_map <- spData::world
library(countrycode)
world_map$country_iso <- countrycode(world_map$name_long, "country.name", "iso3c")
world_map_data <- world_map %>%
  left_join(final_data, by = "country_iso")

choropleth1<-ggplot(world_map_data) +
  geom_sf(aes(fill = Average)) +  # The variable to color by
  scale_fill_gradient(low = "seashell", high = "tomato3")+
  labs(title = "Choropleth of Economic Diversification Index") +
  theme_minimal() +
  theme(legend.position = "bottom")

choropleth1

ggsave(choropleth1, 
       filename = "images/choropleth1.png",
       device = "png")