beauty_of_across3.Rmd

# tidyverse中的across()之美3 {#beauty-of-across3}

```{r beauty-of-across3-1}
library(tidyverse)
```


## 问题

计算每天水分和食物的所占比例， 比如第一天water和food都是10.0，那么各自比例都是50%.
```{r beauty-of-across3-2}
d <- tibble::tribble(
  ~water, ~food,
  10.0,   10.0,
  12.1,   10.3,
  13.5,   19.1,
  17.4,   16.0,
  25.8,   15.6,
  27.4,   19.8
)
d
```

## 传统的方法

传统的方法是，把数据框旋转成长表格，计算所占比例后，再旋转回来
```{r beauty-of-across3-3}
d %>%
  rownames_to_column() %>%
  pivot_longer(
    cols = !rowname
  ) %>%
  group_by(rowname) %>%
  mutate(
    percent = 100 * value / sum(value)
  ) %>%
  ungroup() %>%
  pivot_wider(
    names_from = name,
    values_from = c(value, percent),
    names_glue = "{name}_{.value}"
  )
```

## across的方法

传统的方法，用到基本的dplyr函数，思路很清晰，但有点周折。下面，我列出几个比较新颖的方法，当然这些方法都来源于强大across()函数

### 方法1
```{r beauty-of-across3-4}
d %>%
  mutate(100 * across(.names = "%{.col}") / rowSums(across())) %>%
  ungroup()
```


### 方法2
```{r beauty-of-across3-5}
rowPercent <- function(df) {
  df / rowSums(df) * 100
}

d %>%
  mutate(rowPercent(across(.names = "%{.col}")))
```



### 方法3
```{r beauty-of-across3-6}
d %>%
  rowwise() %>%
  mutate(100 * across(.names = "%{.col}") / sum(c_across())) %>%
  ungroup()
```




### 方法4
```{r beauty-of-across3-7}
scale <- function(x) {
  100 * x / sum(x, na.rm = TRUE)
}

d %>%
  rowwise() %>%
  mutate(
    scale(across(.names = "%{.col}"))
  )
```



### 方法5
```{r beauty-of-across3-8}
d %>%
  rowwise() %>% 
  mutate(100 * proportions(across(.names = "%{.col}")))
```





```{r beauty-of-across3-98, echo = F}
# remove the objects
# ls() %>% stringr::str_flatten(collapse = ", ")

rm(d, scale, rowPercent)
```



```{r beauty-of-across3-99, echo = F, message = F, warning = F, results = "hide"}
pacman::p_unload(pacman::p_loaded(), character.only = TRUE)
```