是否有更优雅的解决方案可将列添加到多个数据框中

时间：2018-08-16 16:08:53

标签： r dplyr data.table apply

我一直在绞尽脑汁想弄清楚如何减少所需的代码量

  • 将数据帧列表中的一列从整数更改为字符
  • 为所有数据帧设置一个data.table键
  • 根据数据帧名称将一列常量添加到一组数据帧中

这是我的代码。我知道必须有更好的方法。

将列从整数修改为字符

MERGED1996.97.PP$OPEID<-as.character(MERGED1996.97.PP$OPEID)
MERGED1997.98.PP$OPEID<-as.character(MERGED1997.98.PP$OPEID)
MERGED1998.99.PP$OPEID<-as.character(MERGED1998.99.PP$OPEID)
MERGED1999.00.PP$OPEID<-as.character(MERGED1999.00.PP$OPEID)
MERGED2000.01.PP$OPEID<-as.character(MERGED2000.01.PP$OPEID)
MERGED2001.02.PP$OPEID<-as.character(MERGED2001.02.PP$OPEID)
MERGED2002.03.PP$OPEID<-as.character(MERGED2002.03.PP$OPEID)
MERGED2003.04.PP$OPEID<-as.character(MERGED2003.04.PP$OPEID)
MERGED2004.05.PP$OPEID<-as.character(MERGED2004.05.PP$OPEID)
MERGED2005.06.PP$OPEID<-as.character(MERGED2005.06.PP$OPEID)
MERGED2006.07.PP$OPEID<-as.character(MERGED2006.07.PP$OPEID)
MERGED2007.08.PP$OPEID<-as.character(MERGED2007.08.PP$OPEID)
MERGED2008.09.PP$OPEID<-as.character(MERGED2008.09.PP$OPEID)
MERGED2009.10.PP$OPEID<-as.character(MERGED2009.10.PP$OPEID)
MERGED2010.11.PP$OPEID<-as.character(MERGED2010.11.PP$OPEID)
MERGED2011.12.PP$OPEID<-as.character(MERGED2011.12.PP$OPEID)
MERGED2012.13.PP$OPEID<-as.character(MERGED2012.13.PP$OPEID)
MERGED2013.14.PP$OPEID<-as.character(MERGED2013.14.PP$OPEID)
MERGED2014.15.PP$OPEID<-as.character(MERGED2014.15.PP$OPEID)
MERGED2015.16.PP$OPEID<-as.character(MERGED2015.16.PP$OPEID)

设置键

setkey(MERGED199<-, "OPEas.character(
setkey(MERGED199<-, "OPEas.character(
setkey(MERGED199<-, "OPEID")
setkey(MERGED1999.00.PP, "OPEID")
setkey(MERGED2000.01.PP, "OPEID")
setkey(MERGED2001.02.PP, "OPEID")
setkey(MERGED2002.03.PP, "OPEID")
setkey(MERGED2003.04.PP, "OPEID")
setkey(MERGED2004.05.PP, "OPEID")
setkey(MERGED2005.06.PP, "OPEID")
setkey(MERGED2006.07.PP, "OPEID")
setkey(MERGED2007.08.PP, "OPEID")
setkey(MERGED2008.09.PP, "OPEID")
setkey(MERGED2009.10.PP, "OPEID")
setkey(MERGED2010.11.PP, "OPEID")
setkey(MERGED2011.12.PP, "OPEID")
setkey(MERGED2012.13.PP, "OPEID")
setkey(MERGED2013.14.PP, "OPEID")
setkey(MERGED2014.15.PP, "OPEID")
setkey(MERGED2015.16.PP, "OPEID")

添加列

MERGED1996.97.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED1996.97.PP))), 9,12)]
MERGED1997.98.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED1997.98.PP))), 9,12)]
MERGED1998.99.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED1998.99.PP))), 9,12)]
MERGED1999.00.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED1999.00.PP))), 9,12)]
MERGED2000.01.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2000.01.PP))), 9,12)]
MERGED2001.02.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2001.02.PP))), 9,12)]
MERGED2002.03.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2002.03.PP))), 9,12)]
MERGED2003.04.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2003.04.PP))), 9,12)]
MERGED2004.05.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2004.05.PP))), 9,12)]
MERGED2005.06.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2005.06.PP))), 9,12)]
MERGED2006.07.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2006.07.PP))), 9,12)]
MERGED2007.08.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2007.08.PP))), 9,12)]
MERGED2008.09.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2008.09.PP))), 9,12)]
MERGED2009.10.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2009.10.PP))), 9,12)]
MERGED2010.11.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2010.11.PP))), 9,12)]
MERGED2011.12.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2011.12.PP))), 9,12)]
MERGED2012.13.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2012.13.PP))), 9,12)]
MERGED2013.14.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2013.14.PP))), 9,12)]
MERGED2014.15.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2014.15.PP))), 9,12)]
MERGED2015.16.PP[, ay :=substr(gsub("\\.","",deparse(substitute(MERGED2015.16.PP))), 9,12)]

chinsoon12中的代码效果很好。现在，我需要更改列名。我添加了

colnames(df)[1] <- "UNITID"

到代码

lapply(dfLs, function(nm) {
df <- get(nm)
setDT(df)[, OPEID := as.character(OPEID)]
setkeyv(df, "OPEID")
colnames(df)[1] <- "UNITID"
df[, ay := substr(gsub("\\.", "", nm), 9, 12)]
})

，但不会修改第一列的名称。任何建议都将不胜感激。

谢谢。

3 个答案:

答案 0 :(得分：0)

以下作品：

sapply(c('data.table', 'ggplot2', 'tictoc'), library, character.only = T, quietly = T)

set.seed(28)
nobv <- 1000 # number of observations per data.table 

l <- lapply(1:5, function(z){
  ind <- sample(x = 1:nrow(diamonds), size = nobv, replace = T)
  return(as.data.table(diamonds[ind,]))
})

# Name all the data.tables
names(l) <- LETTERS[1:length(l)]

cols_to_change <- c('carat', 'z')

# changing cols_to_change to character
lapply(l, function(z){
  z[, (cols_to_change) := lapply(.SD, as.character), .SDcols = cols_to_change]
})

# set key for all data.table
key_by_cols <- c('carat', 'z')
lapply(l, function(z){
  setkeyv(z, key_by_cols)
})

# Add a name column for the data.tables 
lapply(1:length(l), function(z){
  l[[z]]$dtname <<- names(l)[z]
})

答案 1 :(得分：0)

您可能可以执行以下操作：

dfLs <- c("MERGED1996.97.PP","MERGED1997.98.PP","MERGED1998.99.PP","MERGED1999.00.PP",
    "MERGED2000.01.PP","MERGED2001.02.PP","MERGED2002.03.PP","MERGED2003.04.PP",
    "MERGED2004.05.PP","MERGED2005.06.PP","MERGED2006.07.PP","MERGED2007.08.PP",
    "MERGED2008.09.PP","MERGED2009.10.PP","MERGED2010.11.PP","MERGED2011.12.PP",
    "MERGED2012.13.PP","MERGED2013.14.PP","MERGED2014.15.PP","MERGED2015.16.PP")

library(data.table)
lapply(dfLs, function(nm) {
    df <- get(nm)
    setDT(df)[, OPEID := as.character(OPEID)]
    setkeyv(df, "OPEID")
    df[, ay := substr(gsub("\\.", "", nm), 9, 12)]
})

答案 2 :(得分：0)

如果我理解正确，OP必须处理会计数据，这些数据包含在20个单独的数据框中，每个会计年度一个。

OP要求减少执行某些操作所需的代码量。

如果所有数据框确实具有相同的结构（编号，名称，类型和列顺序），则建议将数据绑定到一个大数据对象中。这将大大简化所有操作，包括分组和拆分。

library(data.table)
# create vector of dataframe names
df_names <- sapply(1996:2015, function(yr) sprintf("MERGED%4i.%02i.PP", yr, (yr + 1)%% 100))
# Alternatively, the names can be grabbed from the global environment
df_names <- grep("^MERGED.*PP$", ls(), value = TRUE)
# combine into one data object, prepend id column 
big <- rbindlist(mget(df_names), idcol = "df_name")
big

             df_name     V1  OPEID         V3
              <char> <fctr>  <num>      <num>
 1: MERGED1996.97.PP      C 199601  0.2774292
 2: MERGED1996.97.PP      P 199602  1.0844412
 3: MERGED1997.98.PP      A 199701  0.4291247
 4: MERGED1997.98.PP      F 199702  0.5060559
 5: MERGED1998.99.PP      H 199801 -0.5466319
 6: MERGED1998.99.PP      X 199802 -0.5644520
---                                          
35: MERGED2013.14.PP      H 201301 -1.1088896
36: MERGED2013.14.PP      F 201302 -1.0149620
37: MERGED2014.15.PP      L 201401  0.5630558
38: MERGED2014.15.PP      A 201402  1.6478175
39: MERGED2015.16.PP      F 201501  1.6059096
40: MERGED2015.16.PP      W 201502 -1.1578085

现在我们可以执行以下请求的操作

# change type of OPEID column
big[, OPEID := as.character(OPEID)]
# rename second column (used to be first column before rbindlist())
setnames(BIG, 2, "UNITID")
# append column with accounting year id
big[, ay := substr(gsub("\\.", "", df_name), 9, 12)]
big[]

             df_name UNITID  OPEID         V3     ay
              <char> <fctr>  <num>      <num> <char>
 1: MERGED1996.97.PP      C 199601  0.2774292   9697
 2: MERGED1996.97.PP      P 199602  1.0844412   9697
 3: MERGED1997.98.PP      A 199701  0.4291247   9798
 4: MERGED1997.98.PP      F 199702  0.5060559   9798
 5: MERGED1998.99.PP      H 199801 -0.5466319   9899
 6: MERGED1998.99.PP      X 199802 -0.5644520   9899
---                                                 
35: MERGED2013.14.PP      H 201301 -1.1088896   1314
36: MERGED2013.14.PP      F 201302 -1.0149620   1314
37: MERGED2014.15.PP      L 201401  0.5630558   1415
38: MERGED2014.15.PP      A 201402  1.6478175   1415
39: MERGED2015.16.PP      F 201501  1.6059096   1516
40: MERGED2015.16.PP      W 201502 -1.1578085   1516

请注意，由于setkey()的语法和辅助索引，我不再调用on =，因为联接不再需要它。

可复制的数据

警告：以下代码将在全局环境中创建20个数据框。

set.seed(1234L)
for (yr in 1996:2015) {
  nm <- sprintf("MERGED%4i.%02i.PP", yr, (yr + 1L) %% 100L)
  tmp <- data.frame(V1 = sample(LETTERS, 2L), OPEID = 100* yr + 1:2, V3 = rnorm(2L))
  assign(nm, tmp)
}

ls()

 [1] "big"              "df_names"         "MERGED1996.97.PP" "MERGED1997.98.PP" "MERGED1998.99.PP" "MERGED1999.00.PP"
 [7] "MERGED2000.01.PP" "MERGED2001.02.PP" "MERGED2002.03.PP" "MERGED2003.04.PP" "MERGED2004.05.PP" "MERGED2005.06.PP"
[13] "MERGED2006.07.PP" "MERGED2007.08.PP" "MERGED2008.09.PP" "MERGED2009.10.PP" "MERGED2010.11.PP" "MERGED2011.12.PP"
[19] "MERGED2012.13.PP" "MERGED2013.14.PP" "MERGED2014.15.PP" "MERGED2015.16.PP" "nm"               "tmp"             
[25] "yr"
相关问题
最新问题