Question

我正在尝试使用R为大约4000个城市生成单独的csv文件。为此我编写了一个小的for循环。

所有城市数据都存储在一个称为citys_files的数据框中。在for循环中，我将此合并数据帧拆分为每个城市的小型临时数据帧，并将此临时数据帧写入csv文件。

cities.list <-  unique(cities_files$city_name)

for( cities.list  in cities_files$city_name )   {

tmp <- subset(cities_files,city_name == cities.list) 
cit.name <- unique(tmp$city_name)
fn = paste(paste(cit.name) ,".csv", sep = '')     
write.csv(tmp,fn ,row.names = FALSE)
}

但是，此for循环大约需要3个小时才能生成这4000个文件。我可以使用for循环以外的其他方式来获得所需的结果吗？还是有更好的方法来优化此代码？我想减少生成这些文件所需的时间。

Answer 1

您正在遍历所有观察结果。您可能需要考虑一种data.table方法。

更新
正如@MichaelChirico所指出的，应使用data.table::split方法。

library(data.table)

cities_files <- data.frame(bar = c(1:20000), 
                        city_name  = rep(paste0("city ", 1:200), 100), 
                        foo = c(1:20000))


microbenchmark::microbenchmark(
    khaynes = {
        # library(data.table)
        # Set the data.frame as a data.table
        cities_files_dt <- data.table(cities_files)
        lapply(unique(cities_files_dt[, city_name]), function(city) 
            fwrite(x = subset(cities_files_dt, city_name  == city), 
            file = paste0(city, ".csv")))

    },
    MichaelChirico = {
        cities_files_dt <- data.table(cities_files)
        list_dt <- split(cities_files_dt, cities_files_dt$city_name)
        for(i in 1:length(list_dt)) {
            fwrite(list_dt[[i]], paste0(names(list_dt[i]), ".csv"))
        }
    },
    times = 5
)
# Unit: milliseconds
#       expr      min       lq     mean   median       uq      max neval
#    khaynes 661.0689 680.6768 698.2449 683.1407 719.8056 746.5323     5
#    MichaelChirico 452.0800 456.5777 499.2832 458.0174 517.4398 612.3011     5

使用data.table软件包应大大加快处理时间：

library(data.table)

# Create a dummy data.frame
cities_files <- data.frame(bar = c(1, 1, 2, 3, 3), 
                        city_name  = c("city a", "city a", "city b", "city c", "city c"), 
                        foo = c(20, 14, 40, 50, 60))

# Set the data.frame as a data.table
setDT(cities_files)

lapply(unique(cities_files[, city_name ]), function(city) 
    fwrite(x = subset(cities_files, city_name  == city), 
    file = paste0(city, ".csv")))

有没有比使用for循环更简单的方法来使用R生成关于4000个城市的csv文件？

1 个答案: