Data_exploration_Bibata.Rmd

---
title: "Data Exporation"
output: html_document
date: "2023-06-21"
---

```{r Setup}

library(tidyverse)
library(here)
library(lubridate)
here()
city_day_data <- read_csv("city_day_agg_cleaned.csv.gz")

country_day_data <- read_csv("country_day_agg_cleaned.csv.gz")
```

Looking at the data we have two tables with the following features:

#### city_day_data (Features)

- country
- countryCode
- city_id
- date
- parameter
- mean

#### country_data_data (Features)

- countryCode
- date
- parameter
- mean

```{r Data Preparation}

wider_city_data <- city_day_data %>% 
  pivot_wider(names_from=parameter, values_from=mean) %>% 
  select(-c(no2, o3)) %>% 
  na.omit() %>% 
  rename(city_average_pm25=pm25)

wider_country_data <- country_day_data %>% 
  pivot_wider(names_from=parameter, values_from=mean) %>% 
  select(-c(no2, o3)) %>% 
  na.omit() %>% 
  rename(country_average_pm25=pm25)

  wider_city_data %>% 
  left_join(wider_country_data, by=c("countryCode", "date")) %>% 
  mutate(year_period=as.factor(ifelse(year(date) < 2020, "3-year", "2020"))) %>% 
  group_by(countryCode, city_id, year_period) %>% 
  summarize(mean_pm25=mean(city_average_pm25)) %>% 
  group_by(year_period) %>% 
  mutate(median_2020=median(mean_pm25)) %>% 
  ggplot(aes(x=mean_pm25, color=year_period)) + 
  geom_density(aes(fill=year_period, alpha=0.5)) +
  geom_vline(aes(xintercept = median_2020, color=year_period))
  
  
  
  
  


```