00-exploratory-analysis.qmd

---
title: "Exploratory data analysis"
author: "Diego Villa"
format: html
editor: source
---

```{r}
#| warning: false
#| message: false
library(dplyr)
library(ggplot2)
```

## Spatio-temporal maps

```{r}
file_path = "data/processed/dengue-climate.csv"
dengue_climate = readr::read_csv(file_path)
glimpse(dengue_climate)
```

```{r}
dengue = dengue_climate |> 
  # mutate(incidence = 100000 * (cases / population)) |> 
  group_by(ubigeo, year) |> 
  # summarise(incidence = sum(incidence, na.rm = TRUE), .groups = "drop")
  summarise(cases = sum(cases, na.rm = TRUE), .groups = "drop")
```

```{r}
library(sf)
dengue_climate_geom = readRDS("data/processed/dengue-climate-geom.rds")
```

```{r}
dengue_geom = dengue |> 
  left_join(dengue_climate_geom, by = "ubigeo") |> 
  st_as_sf()
```

```{r}
#| fig-width: 8
#| fig-height: 9
dengue_years_plot = ggplot(dengue_geom) +
  facet_wrap(~year, ncol = 3) +
  geom_sf(aes(fill = log(cases + 1)), color = NA) +
  theme_void() +
  theme(legend.position = "bottom") +
  scale_fill_viridis_c(name = "Cumulative\ncases")

dengue_years_plot
```

```{r}
#| eval: false
ggsave("figures/dengue-cases-years.png", dengue_years_plot, width = 8, height = 9)
```