将行最大值从长到宽投射

Question

我有一个DT如下（谁能解释一下为什么这里的fread不能正常工作？）：

library(data.table)
DT <- fread("
pty_n  pty  pev1    vot1    vv1     ivv1    to1         pv1       ctr_n     yr
A      236  6567389 5759215 5554662 204553  876941383   2743064   Argentina 1983
B      41   6567389 5759215 5554662 204553  876941383   56987     Argentina 1983
C      237  6567389 5759215 5554662 204553  876941383   62659     Argentina 1983
D      207  6567389 5759215 5554662 204553  876941383   587       Argentina 1983
E      125  6567389 5759215 5554662 204553  876941383   3728      Argentina 1983",
header=TRUE)

我想按如下方式更改 DT 的格式：

1) 对于pv1最高的两个pty_n，我想要党名。

ctr_n     yr   First_Party  Second_Party
Argentina 1983 A            C

2）我想按比例加上他们的得票份额。

ctr_n     yr   First_Party  Second_Party  First_Party_Votes                   First_Party_Votes
Argentina 1983 A            C             pv1/vot1 = 2743064/5759215 = 0.476  0.011 (rounded from 0.0108)

我正在考虑做如下事情：

# Selecting the row with the maximum amount of votes (But how do I do the second max?)
DT[, .SD[max(pv1)==prop_votes ], on=.(ctr_n, yr)]
# Calculating the proportion for the remaining rows
DT[, prop_votes := pv1/vot1]
# Recasting, here I get stuck as well
DT <- reshape2::dcast(DT, ctr_n + yr ~ pty_n + pty + prop_votes + ?

有人可以帮我继续吗？

Answer 1

是吗？

DT[order(-pty)[1:2]
   ][, .(ctr_n  = ctr_n[1], 
         First_Party= pty_n[1], 
         Second_Party= pty_n[2], 
         yr = yr[1],
         First_Party_Votes = pv1[1] / vot1[1]
         )]

       ctr_n First_Party Second_Party   yr First_Party_Votes
1: Argentina           C            A 1983        0.01087978

Answer 2

我的尝试基于@sindri_baldur的回答

library(data.table)
DT <- fread("
pty_n  pty  pev1    vot1    vv1     ivv1    to1         pv1       ctr_n     yr
A      236  6567389 5759215 5554662 204553  876941383   2743064   Argentina 1983
B      41   6567389 5759215 5554662 204553  876941383   56987     Argentina 1983
C      237  6567389 5759215 5554662 204553  876941383   62659     Argentina 1983
D      207  6567389 5759215 5554662 204553  876941383   587       Argentina 1983
E      125  6567389 5759215 5554662 204553  876941383   3728      Argentina 1983
A      236  6567389 5759215 5554662 204553  876941383   56987  Argentina 1984
B      41   6567389 5759215 5554662 204553  876941383   56987     Argentina 1984
C      237  6567389 5759215 5554662 204553  876941383   62659     Argentina 1984
D      207  6567389 5759215 5554662 204553  876941383   2743064       Argentina 1984
E      125  6567389 5759215 5554662 204553  876941383   3728      Argentina 1984",
            header=TRUE)

DT[order(ctr_n, yr, -pv1)
   ][, .(First_Party= pty_n[1], 
         Second_Party= pty_n[2], 
         First_Party_Votes = pv1[1] / vot1[1],
         Second_Party_Votes = pv1[2] / vot1[1]
   ), .(ctr_n, yr)]

       ctr_n   yr First_Party Second_Party First_Party_Votes Second_Party_Votes
1: Argentina 1983           A            C         0.4762913         0.01087978
2: Argentina 1984           D            C         0.4762913         0.01087978

Answer 3

不太确定这是不是你想要的，但你可以试试

h <- head(DT[order(DT$pv1,decreasing = T),],2)
df <- data.frame(t(c(unlist(Map(as.character,with(h,list(unique(ctr_n),unique(yr),pty_n, round(pv1/vot1,3))))))))
colnames(df) <- c("ctr_n","yr","First_Party","Second_Party","First_Party_Votes","Second_Party_Votes")

屈服

> df
      ctr_n   yr First_Party Second_Party First_Party_Votes Second_Party_Votes
1 Argentina 1983           A            C             0.476              0.011

将行最大值从长到宽投射

Casting the row maxima from long to wide

r

fread

data.table

dcast