R：循环在 R 中创建新的数据帧

Question

我正在尝试创建一个循环，该循环为 VCS 站创建多个文件，这些文件根据其站名命名。下面是为一个站点执行此操作的代码，我正在尝试将其变成一个循环，以便可以为 68 个站点执行此操作。 （即，如果我正在复制和粘贴，我将用不同的电台名称替换 P205187，例如 P205200）。我在一个名为 VCS.Sites 的数据框中有单独的电台名称（例如 P205187）。谁能指出我正确的方向？新的 R 用户在这里，我被困住了！

P205187 <- VCSrawdata[VCSrawdata$Network_ID=="P205187",]  #create a file for VCS station P205187
  #clean up after subset
  P205187$Network_ID <- factor(P205187$Network_ID)


# create annual file for VCS station P205187
P205187_annual <- group_by(P205187,Year,DESCRIPTION)
P205187_annual <- summarise(P205187_annual,Sum_Annual = sum(Value), Mean_Annual = mean(Value), CountDays=n())

# create monthly file for VCS station P205187
P205187_monthly <- group_by(P205187,Year, Month,DESCRIPTION)
P205187_monthly <- summarise(P205187_monthly,Sum_Monthly = sum(Value),Mean_monthly = mean(Value),CountDays=n())

Answer 1

你可以用一个 lapply 循环做得很好。像这样的：

list_of_ids <- c("List be here")

monthly <- function(id){
  P205187 <- VCSrawdata[VCSrawdata$Network_ID==id,]  #create a file for VCS station P205187
  #clean up after subset
  P205187$Network_ID <- factor(P205187$Network_ID)
  
  
  # create annual file for VCS station P205187
  P205187_annual <- group_by(P205187,Year,DESCRIPTION)
  P205187_annual <- summarise(P205187_annual,Sum_Annual = sum(Value), Mean_Annual = mean(Value), CountDays=n())
  
  # create monthly file for VCS station P205187
  P205187_monthly <- group_by(P205187,Year, Month,DESCRIPTION)
  P205187_monthly <- summarise(P205187_monthly,Sum_Monthly = sum(Value),Mean_monthly = mean(Value),CountDays=n())
  
  return(P205187_monthly)
}

monthlies <- lapply(list_of_ids, monthly)

Answer 2

听起来这是为了写 csvs。我们可以使用dplyr 中的group_map 循环遍历所有站点并写入csvs。

library(dplyr)

VCSrawdata %>%
    group_by(Network_ID) %>%
    group_walk(~ {
        .x%>%
            group_by(Year, DESCRIPTION) %>%
            summarize(sum_annual = sum(Value),
                      mean_annual = mean(Value),
                      countDays = n())%>%
            write.csv(file = paste0(.y$Network_ID, "_annual_csv"))
        
        .x%>%
            group_by(Year, Month, DESCRIPTION) %>%
            summarize(sum_month = sum(Value),
                      mean_month = mean(Value),
                      countDays = n())%>%
            write.csv(file = paste0(.y$Network_ID, "_month_csv")) 
    }
    )

注意事项：

1. .x 指的是被Network_ID 分割的分组标题
2. .y 指的是分组。在这种情况下，我们只有Network_ID。

Answer 3

只需在定义的方法中概括您的流程，然后将站点名称作为参数传递给循环或 apply 函数以遍历站点。使用这种方法，您可以避免许多个单独的对象泛滥全局环境，而是使用一个单个命名的许多底层元素列表来更好地序列化和组织。

summarize_stations <- function(station_name) { 

   tmp_df <- VCSrawdata[VCSrawdata$Network_ID==station_name,] 
   tmp_df$Network_ID <- factor(tmp_df$Network_ID) 

   # create annual file for VCS station
   tmp_annual <- summarise(group_by(tmp,Year,DESCRIPTION),
                           Sum_Annual = sum(Value), 
                           Mean_Annual = mean(Value), 
                           CountDays=n()) 

   # create monthly file for VCS station
   tmp_monthly <- summarise(group_by(tmp, Year, Month,DESCRIPTION),
                            Sum_Annual = sum(Value), 
                            Mean_Annual = mean(Value), 
                            CountDays=n())

   # RETURN NAMED LIST OF BOTH AGGREGATIONS
   return(list(annual=tmp_annual, monthly=tmp_monthly))
}

station_list <- sapply(VCS.Sites$station_names, summarize_stations, simplify=FALSE)


# ACCESS UNDERLYING ELEMENTS
station_list$P205187$annual
station_list$P205187$monthly
...

---

您甚至可以使用by（tapply 的面向对象的包装器）通过Network_ID 将VCSrawdata 子集化（假设它包括您需要的所有站点）。为此，请稍微调整函数以接收数据帧作为参数，从而允许您跳过子集行。

summarize_stations <- function(tmp_df) { 

  # REMOVE SUBSET LINE
  # tmp_df <- VCSrawdata[VCSrawdata$Network_ID=="P205187",] 
  
  ...keep same code as above    
}

station_list <- by(VCSrawdata, VCSrawdata$Network_ID, FUN=summarize_stations)


# ACCESS UNDERLYING ELEMENTS
station_list$P205187$annual
station_list$P205187$monthly
...