eda_migration.Rmd

# 探索性数据分析-移民缺口 {#eda-migration}


## 引言

今天看到一张图，觉得很不错，简单清晰。

```{r, out.width = '90%', echo = FALSE}
knitr::include_graphics(path = "images/migration.png")
```


数据是公开的，因此不难找到，我是在[这里](https://www.cato.org/blog/where-did-24-million-48-million-missing-immigrants-go)图中获取。

**先观察这张图想表达的意思：**

- 蓝色的是历年移民人口真实数据

- 依据前6个点（2011年到2016年）建立线性模型，并依此预测后5个点（2016到2021年）的情况，从而得到黄色的直线

- 预测情况与实际情况的差，得到缺口总数210万



## 开始

```{r, message=FALSE, warning=FALSE}
library(tidyverse)
library(modelr)

df <- tibble::tribble(
  ~year,   ~num,
  2011,  795300,
  2012,  858740,
  2013,  849730,
  2014,  945640,
  2015, 1060000,
  2016, 1065000,
  2017,  948390,
  2018,  719870,
  2019,  568540,
  2020,  477030
) %>%
  mutate(num = num / 1000000)

df
```



### 线性回归模型

依据2011到2016年的数据，建立线性模型

```{r, message=FALSE, warning=FALSE}
mod <- df %>%
  filter(year <= 2016) %>%
  lm(num ~ 1 + year, data = .)
```


### 预测

根据线性模型预测2016年到2020的情况

```{r, message=FALSE, warning=FALSE}
pred_df <- tibble(
  year = seq(2016, 2020, by = 1)
) %>%
  modelr::add_predictions(model = mod)

pred_df
```

合并成新的数据框
```{r}
combine_df <- df %>%
  left_join(pred_df, by = "year")
combine_df
```



```{r, eval=FALSE}
# 一个等价的方法
df %>%
  modelr::add_predictions(model = mod) %>%
  mutate(pred = if_else(year < 2016, NA_real_, pred))
```


### 可视化

- 基本绘图，画折线图和散点图
```{r}
combine_df %>%
  ggplot(aes(x = year, y = num)) +
  geom_point() +
  geom_line() +
  geom_line(aes(y = pred)) +
  geom_point(aes(y = pred))
```



- 调整坐标和配色

```{r}
mycolor <- c("#3D81E0", "#f4a261")

scales::show_col(mycolor)
```



```{r}
combine_df %>%
  ggplot(aes(x = year, y = num)) +
  geom_point(size = 4, fill = "#3D81E0", color = "#3D81E0", shape = 23) +
  geom_line(size = 2, color = "#3D81E0") +
  geom_line(aes(y = pred), size = 2, color = "#f4a261") +
  geom_point(aes(y = pred), size = 4, fill = "#f4a261", color = "#f4a261", shape = 23) +
  labs(
    title = "Figure2: Contribution of international migration to population growth",
    subtitle = "millions per year",
    caption = "Source: Census Bureau, J.P. Morgan",
    x = NULL,
    y = NULL
  ) +
  scale_y_continuous(
    limits = c(0, 1.5),
    breaks = seq(0, 1.5, by = 0.25),
    expand = c(0, 0)
  ) +
  scale_x_continuous(
    limits = c(2011, 2021),
    breaks = seq(2011, 2021, by = 1),
    expand = c(0.1, 0)
  ) +
  theme_classic(base_size = 14) +
  theme(
    legend.position = "none",
    plot.title.position = 'plot',
    plot.caption = element_text(hjust = 0)
  )
```



- 添加标注

```{r}
arrows <- tibble::tribble(
  ~x1, ~y1, ~x2, ~y2, ~color,
  2016, 0.54, 2017.5, 0.75, "a",
  2017, 1.42, 2018.0, 1.25, "b"
)



combine_df %>%
  ggplot(aes(x = year, y = num)) +
  geom_point(size = 4, fill = "#3D81E0", color = "#3D81E0", shape = 23) +
  geom_line(size = 2, color = "#3D81E0") +
  geom_line(aes(y = pred), size = 2, color = "#f4a261") +
  geom_point(aes(y = pred), size = 4, fill = "#f4a261", color = "#f4a261", shape = 23) +
  geom_ribbon(
    aes(ymin = num, ymax = pred),
    fill = "orange",
    alpha = 0.2
  ) +
  geom_segment(
    data = arrows,
    aes(x = x1, y = y1, xend = x2, yend = y2, color = color),
    arrow = arrow(length = unit(0.15, "inch")), size = 1.5
  ) +
  annotate("text",
    x = c(2017, 2016, 2021), y = c(1.47, 0.5, 0.9),
    size = 6, face = "bold",
    label = c("Pre-2017 trend", "Actual", "Shortfall:\n 2.1 million")
  ) +
  labs(
    title = "Figure2: Contribution of international migration to population growth",
    subtitle = "millions per year",
    caption = "Source: Census Bureau, J.P. Morgan",
    x = NULL,
    y = NULL
  ) +
  scale_y_continuous(
    limits = c(0, 1.5),
    breaks = seq(0, 1.5, by = 0.25),
    expand = c(0, 0)
  ) +
  scale_x_continuous(
    limits = c(2011, 2021),
    breaks = seq(2011, 2021, by = 1),
    expand = c(0.1, 0)
  ) +
  scale_color_manual(
    values = c(a = "#3D81E0", b = "#f4a261")
  ) +
  theme_classic(base_size = 14) +
  theme(
    legend.position = "none",
    plot.title.position = 'plot',
    plot.caption = element_text(hjust = 0)
  )
```


- 保存
```{r, eval = FALSE}
ggsave("migration.pdf", width = 8, height = 5)
```




```{r, echo = F}
# remove the objects
# rm(list=ls())
rm(arrows, combine_df, mycolor, df, pred_df, mod)
```

```{r, echo = F, message = F, warning = F, results = "hide"}
pacman::p_unload(pacman::p_loaded(), character.only = TRUE)
```