通过组合所有数据帧中的类似条目,从列表中形成新的数据帧

时间:2015-03-11 16:36:15

标签: r list lapply cbind

我有一个包含许多数据帧的列表(示例如下)。

    G100=structure(list(Return.Period = structure(c(4L, 6L, 2L, 3L, 5L, 
        1L), .Label = c("100yrs", "10yrs", "20yrs", "2yrs", "50yrs", 
        "5yrs"), class = "factor"), X95..lower.CI = c(54.3488053692529, 
        73.33363378538, 84.0868168935697, 91.6191228597281, 96.3360349026068, 
        95.4278817251266), Estimate = c(61.6857930414643, 84.8210149260708, 
        101.483909733627, 118.735593472652, 143.33257990536, 163.806035490329
        ), X95..upper.CI = c(69.0227807136758, 96.3083960667617, 118.881002573685, 
        145.852064085577, 190.329124908114, 232.18418925553)), .Names = c("Return.Period", 
        "X95..lower.CI", "Estimate", "X95..upper.CI"), row.names = c(NA, 
        -6L), class = "data.frame")

G101<-G100 # just for illustration

mylist=list(G100,G101) # there 100 of these with differet codes

names(mylist)代表“SITE”。从每个数据帧,我想采用“估计”并形成一个新的数据框,看起来像这样(不完全是因为所有dfs的值都不相同): 估计<-

SITE    X2yrs    X5yrs   X10yrs   X20yrs   X50yrs X100yrs
G100 61.68579 84.82101 101.4839 118.7356 143.3326 163.806
G101 61.68579 84.82101 101.4839 118.7356 143.3326 163.806

请注意,SITEmylist中的数据框名称相同。

"X95..lower.CI""X95..upper.CI"执行相同操作。

因此,我将使用上述布局最终得到3个数据框"Estimate""X95..lower.CI""X95..upper.CI".

#lapply, rbindlist,cbind and others can do but how?

建议请。

1 个答案:

答案 0 :(得分:0)

只需使用for循环添加名称即可。这可能是一种花哨的*apply方式,但for易于使用,记忆和理解。

首先添加名称:

names(mylist) = paste0("G", seq(from = 100, by = 1, length.out = length(mylist)))

像以前一样添加SITE列:

for (i in seq_along(mylist)) {
    mylist[[i]]$SITE = names(mylist)[i]
}

合并数据框:

由于您拥有大量数据框或者它们相当大,因此请使用dplyr::rbind_all来提高速度。 (在基数R中,do.call(rbind, mylist)会起作用,但速度会慢一些。)

library(dplyr)
combined = bind_rows(mylist)

(较早版本的dplyr可以使用rbind_all代替bind_rows,但很快就会弃用:https://github.com/hadley/dplyr/issues/803)。)

将Estimate和CI列从long转换为wide。

tidyr这很容易,但reshape2::dcast的工作方式类似:

library(tidyr)
Estimate = combined %>% select(SITE, Return.Period, Estimate) %>%
    spread(key = Return.Period, value = Estimate)
head(Estimate)
# Source: local data frame [2 x 7]
#
#   SITE  100yrs    10yrs    20yrs     2yrs    50yrs     5yrs
# 1 G100 163.806 101.4839 118.7356 61.68579 143.3326 84.82101
# 2 G101 163.806 101.4839 118.7356 61.68579 143.3326 84.82101    

Lower95 = combined %>% select(SITE, Return.Period, X95..lower.CI) %>%
    spread(key = Return.Period, value = X95..lower.CI)
head(Lower95)
# Source: local data frame [2 x 7]
#
#   SITE   100yrs    10yrs    20yrs     2yrs    50yrs     5yrs
# 1 G100 95.42788 84.08682 91.61912 54.34881 96.33603 73.33363
# 2 G101 95.42788 84.08682 91.61912 54.34881 96.33603 73.33363

您可能希望按字母顺序重新排序列。

  

为``&#34; X95..upper.CI&#34;`做同样的事。

仍留作读者的练习。