Question

避免下面使用的for循环的最佳方法是什么？它们遍历“状态”，使用它们对数据进行子集化以绘制标题并标记标题。也许apply函数适用于用户定义的函数，但我不确定它会是什么，它也避免了for循环。

# Attach packages
library(ggplot2)
library(dplyr)
library(gridExtra)

# Create data
set.seed(123)
states <- c("s1","s2","s3")
d <- data.frame(Year = sample(2010:2016, 1000, replace=T), 
           Month = sample(1:12, 1000, replace=T), 
           DepartureState = sample(states, 1000, replace=T), 
           DestinationState = sample(states, 1000, replace=T), 
           Price = sample(5000:8000, 1000, replace=T), 
           Cost = sample(2000:3000, 1000, replace=T))

# Apply grouping
dg <- d %>%
  group_by( Year, Month, DepState = DepartureState, DestState = DestinationState ) %>% 
  summarise( sumPrice = sum(Price), sumCost = sum(Cost), diff = sumPrice-sumCost, vol = n() )

# Add date column
dg$date <- as.POSIXct(paste(dg$Year, dg$Month, "01", sep = "-"))      

# Do things, e.g. subset and plot, for each combination of DepState-DestState pairs
for ( depState in states ) { 
  for ( destState in states ) {
    dgcut <- dg[dg$DepState == depState & dg$DestState == destState, ]

    description <- paste0(depState," to ", destState)
    plotname <- paste0(depState,"_",destState)
    #png(filename=paste0(plotname,".png"))
    p1 <- ggplot(dgcut, aes(x=date, y=sumPrice)) + geom_line() 
    p2 <- ggplot(dgcut, aes(x=date, y=sumCost)) + geom_line()
    p3 <- ggplot(dgcut, aes(x=date, y=diff)) + geom_line()
    p4 <- ggplot(dgcut, aes(x=date, y=vol)) + geom_line()
    grid.arrange(p1,p2,p3,p4,ncol=1,top = description) # from gridExtra
    #dev.off()
  }
}

注意，我只是试图避免for循环，因为我理解它在像R这样的矢量化语言中被认为是不好的做法（或次优）。如果对于这种情况不是这样，请告诉我！

Answer 1

链接lapply代替for循环？

doThings <- function(depState,destState) {
    dgcut <- dg[dg$DepState == depState & dg$DestState == destState, ]

    description <- paste0(depState," to ", destState)
    plotname <- paste0(depState,"_",destState)
    #png(filename=paste0(plotname,".png"))
    p1 <- ggplot(dgcut, aes(x=date, y=sumPrice)) + geom_line() 
    p2 <- ggplot(dgcut, aes(x=date, y=sumCost)) + geom_line()
    p3 <- ggplot(dgcut, aes(x=date, y=diff)) + geom_line()
    p4 <- ggplot(dgcut, aes(x=date, y=vol)) + geom_line()
    grid.arrange(p1,p2,p3,p4,ncol=1,top = description) # from gridExtra
    #dev.off()
}

lapply(states,function(x) lapply(states,doThings,destState = x))

在r中进行子集化和绘图时避免循环

1 个答案: