专家,寻找一些建议与下面的R数据框我需要建立一个特定的城市内的每个区域的关系。R dataframe创建一对一的关系
输入:
mydf = data.frame(City = c("LA", "LA", "LA", "NYC", "NYC"),
Zone = c("A1", "A2", "A3", "B1", "B2"))
预期输出:
专家,寻找一些建议与下面的R数据框我需要建立一个特定的城市内的每个区域的关系。R dataframe创建一对一的关系
输入:
mydf = data.frame(City = c("LA", "LA", "LA", "NYC", "NYC"),
Zone = c("A1", "A2", "A3", "B1", "B2"))
预期输出:
下面是定义组合&功能的tidyverse方法适用于每个城市的区域:
library(dplyr); library(tidyr); library(purrr)
generate_combinations <- function(data){
zone <- data %>% select(Zone) %>% unlist()
combinations <- expand.grid(Zone_1 = zone, Zone_2 = zone) # generate all combinations
combinations <- combinations %>%
filter(!(Zone_1 == Zone_2)) %>% # remove invalid combinations
mutate_all(as.character)
return(combinations)
}
mydf <- mydf %>%
nest(Zone) %>%
mutate(data = map(data, generate_combinations)) %>%
unnest()
> mydf
City Zone_1 Zone_2
1 LA A2 A1
2 LA A3 A1
3 LA A1 A2
4 LA A3 A2
5 LA A1 A3
6 LA A2 A3
7 NYC B2 B1
8 NYC B1 B2
# if City info is no longer needed
mydf <- mydf %>% select(-City)
数据:
mydf = data.frame(City = c("LA", "LA", "LA", "NYC", "NYC"),
Zone = c("A1", "A2", "A3", "B1", "B2"),
stringsAsFactors = F)
完美。谢谢!! – Kg211
这几乎是肯定不会做的事情最有效的方式,但它会工作,这是几乎可读。
library(tidyverse)
library(magrittr)
output <- mydf %>%
split(., f=mydf[, "City"]) %>% # Split into data.frames by "City"
sapply(., function(x) use_series(x, Zone)) %>% # Extract zones
sapply(combn, 2) %>% # Find all combinations of size 2
do.call("cbind", .) %>% # Combine them into a data frame
t %>%
as.data.frame %>%
rbind(., data.frame(V1=.$V2, V2=.$V1)) # Add it to the inverse, to get all possible combinations
colnames(output) <- c("Zone_1", "Zone_2") # Rename columns
output
Zone_1 Zone_2
1 A1 A2
2 A1 A3
3 A2 A3
4 B1 B2
5 A2 A1
6 A3 A1
7 A3 A2
8 B2 B1
完美。如预期。谢谢!! – Kg211
是否存在的第2行中的错字输出表? (A1,A2)重复两次。应该是(A1,A3)而不是? –
是的,你是对的。它应该是(A1,A3) – Kg211
你几乎可以通过使用'melt(crossprod(table(mydf)))'到达那里,但为了得到预期的结果,你可以使用'temp < - crossprod(table(mydf)); diag(temp)< - NA; r < - reshape2 :: melt(temp,na.rm = TRUE); r [r $ value == 1,]' – user20650