在 df 中自动改变多列

Question

我想从列列表中自动生成数据框中的交互。 我有这个例子..

library(tidyverse)
library(nycflights13)

df<- nycflights13::weather
interactions <- c("hour","temp","dewp","humid","wind_dir","wind_speed")
df<- df %>% mutate (hour_temp= hour*temp, hour_dewp= hour*dewp)

现在我试图弄清楚如何自动生成所有列（例如继续使用 temp dewp、temp hum ...）

有任何想法吗？

Answer 1

我们可以使用combn来获取“interactions”中的所有组合，一次取 2 个，在combn指定FUN以select这些列， transmute乘积 ( * ) 并将它们与原始数据集绑定

library(dplyr)
library(purrr)
library(stringr)
out <- combn(interactions, 2, function(x) df %>%
       dplyr::select(x) %>%
       transmute(!! str_c(x, collapse="_"):= reduce(., `*`)),
          simplify = FALSE) %>% 
   bind_cols %>% 
   bind_cols(df, .)

- 检查输出

names(out)
#[1] "origin"              "year"                "month"               "day"                 "hour"               
#[6] "temp"                "dewp"                "humid"               "wind_dir"            "wind_speed"         
#[11] "wind_gust"           "precip"              "pressure"            "visib"               "time_hour"          
#[16] "hour_temp"           "hour_dewp"           "hour_humid"          "hour_wind_dir"       "hour_wind_speed"    
#[21] "temp_dewp"           "temp_humid"          "temp_wind_dir"       "temp_wind_speed"     "dewp_humid"         
#[26] "dewp_wind_dir"       "dewp_wind_speed"     "humid_wind_dir"      "humid_wind_speed"    "wind_dir_wind_speed"

---

或使用base R

m1 <- combn(interactions, 2)
df[paste(m1[1,], m1[2, ], sep="_")] <- lapply(asplit(m1, 2), 
       function(x) do.call(`*`, df[x]))