---
title: "Use case: migration flows"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Use case: migration flows}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

In this use case we show how to use [`tidycenso`](https://github.com/rOpenSpain/tidycenso) to visualize the most important internal migration flows over the last 10 years.

First download the number of people by province of residence in 2011 and 2020.

```{r}
library(tidycenso)
library(dplyr)

df <- get_censo("per.ppal", c("ID_RESI_2011_N2", "ID_RESI_2020_N2"), "SPERSONAS")

head(df)
```

We then do some data wrangling estimate the net flows, and select the largest ones.

```{r}

df <- 
  df %>% 
  filter(
    ID_RESI_2011_N2 != "No consta",
    ID_RESI_2020_N2 != "No consta",
  ) %>% 
  mutate(across(c(ID_RESI_2020_N2, ID_RESI_2011_N2), ~substr(., 1, 2))) 

entries <- 
  df %>% 
  rename(to = ID_RESI_2020_N2, from = ID_RESI_2011_N2)

exits <- 
  df %>% 
  rename(from = ID_RESI_2020_N2, to = ID_RESI_2011_N2) %>% 
  mutate(SPERSONAS = -SPERSONAS)


net_flows <- 
  rbind(entries, exits) %>% 
  group_by(from, to) %>% 
  summarise(SPERSONAS = sum(SPERSONAS)) %>% 
  filter(
    SPERSONAS > 0,
    !from %in% c("Ex", "No"),
    !to %in% c("Ex", "No")
  ) %>% 
  ungroup() %>% 
  slice_max(SPERSONAS, prop = .10)

```

With the estimated net migration flows, we can now visualize them We'll use [`tidygraph`](https://tidygraph.data-imaginist.com/) and [`ggraph`](https://ggraph.data-imaginist.com/index.html) to create the graph, [`ggplot2`](https://ggplot2.tidyverse.org/) to visualize it and [`mapSpain`](https://ropenspain.github.io/mapSpain/index.html) to download the geometries of the provinces.

```{r}

library(tidygraph)
library(ggraph)
library(mapSpain)
library(ggplot2)
library(sf)

sf_provs <- esp_get_prov() %>%
  select(cpro, ine.prov.name)

nodes <- 
  sf_provs %>% 
  st_centroid() %>% 
  st_coordinates() %>% 
  as.data.frame() %>% 
  mutate(id = sf_provs$cpro)

graph <- tbl_graph(nodes, edges = net_flows, node_key = "id", directed = T)

coord_layout <- 
  create_layout(
    graph = graph,
    layout = "manual", 
    x = nodes$X,
    y = nodes$Y
  )


```


Once we have created the graph and the layout for it, we can use [`ggplot2`](https://ggplot2.tidyverse.org/) to visualise it.
```{r}

p <- 
  ggraph(coord_layout) +
  geom_sf(data = sf_provs, color = "black",  fill = "grey90") +
  geom_edge_arc(
    aes(
      width=SPERSONAS,
      alpha = after_stat(index),
      color = after_stat(index)
    ),
    lineend = "round", n = 200,
    strength = .1) +
  scale_edge_width_continuous(range=c(0.1, 1.5)) +
  scale_edge_color_gradientn(colors = scales::brewer_pal("seq", direction = 1)(5)[3:5]) + 
  scale_edge_alpha(range = c(.05, 1)) +
  theme_void() +
  theme(legend.position = "none")

p

```