R 计算每组的行数,并将结果添加到原始数据帧
假设我有一个R 计算每组的行数,并将结果添加到原始数据帧,r,count,aggregate,r-faq,R,Count,Aggregate,R Faq,假设我有一个data.frame对象: df <- data.frame(name=c('black','black','black','red','red'), type=c('chair','chair','sofa','sofa','plate'), num=c(4,5,12,4,3)) 或者也可以使用plyr,(尽管我不确定如何使用) 但是,如何将结果合并到原始数据框中?结果如下所示: df # name
data.frame
对象:
df <- data.frame(name=c('black','black','black','red','red'),
type=c('chair','chair','sofa','sofa','plate'),
num=c(4,5,12,4,3))
或者也可以使用plyr
,(尽管我不确定如何使用)
但是,如何将结果合并到原始数据框中?结果如下所示:
df
# name type num count
# 1 black chair 4 2
# 2 black chair 5 2
# 3 black sofa 12 1
# 4 red sofa 4 1
# 5 red plate 3 1
# generate vector of 0s
df$count <-0L
# fill it in
split(df$count, df[c("name", "type")]) <- lengths(split(df$num, df[c("name", "type")]))
其中count
现在存储聚合结果
使用plyr
的解决方案也很有意思,不过我想看看如何使用base R实现这一点。您可以这样做:
> ddply(df,.(name,type),transform,count = NROW(piece))
name type num count
1 black chair 4 2
2 black chair 5 2
3 black sofa 12 1
4 red plate 3 1
5 red sofa 4 1
或者更直观地说
> ddply(df,.(name,type),transform,count = length(num))
name type num count
1 black chair 4 2
2 black chair 5 2
3 black sofa 12 1
4 red plate 3 1
5 red sofa 4 1
您可以使用
ave
:
df$count <- ave(df$num, df[,c("name","type")], FUN=length)
df$count使用数据。表
:
library(data.table)
dt = as.data.table(df)
# or coerce to data.table by reference:
# setDT(df)
dt[ , count := .N, by = .(name, type)]
library(dplyr)
df %>%
group_by(name, type) %>%
mutate(count = n())
plyr::ddply(df, .(name, type), transform, count = length(num))
有关预处理的数据。表1.8.2
备选方案,请参阅编辑历史记录
使用dplyr
:
library(data.table)
dt = as.data.table(df)
# or coerce to data.table by reference:
# setDT(df)
dt[ , count := .N, by = .(name, type)]
library(dplyr)
df %>%
group_by(name, type) %>%
mutate(count = n())
plyr::ddply(df, .(name, type), transform, count = length(num))
或者简单地说:
add_count(df, name, type)
使用plyr
:
library(data.table)
dt = as.data.table(df)
# or coerce to data.table by reference:
# setDT(df)
dt[ , count := .N, by = .(name, type)]
library(dplyr)
df %>%
group_by(name, type) %>%
mutate(count = n())
plyr::ddply(df, .(name, type), transform, count = length(num))
另一种概括更多的方式:
df$count <- unsplit(lapply(split(df, df[c("name","type")]), nrow), df[c("name","type")])
df$count基本R
函数aggregate
将使用一行代码获取计数,但将这些计数添加回原始数据。帧似乎需要一些处理
df <- data.frame(name=c('black','black','black','red','red'),
type=c('chair','chair','sofa','sofa','plate'),
num=c(4,5,12,4,3))
df
# name type num
# 1 black chair 4
# 2 black chair 5
# 3 black sofa 12
# 4 red sofa 4
# 5 red plate 3
rows.per.group <- aggregate(rep(1, length(paste0(df$name, df$type))),
by=list(df$name, df$type), sum)
rows.per.group
# Group.1 Group.2 x
# 1 black chair 2
# 2 red plate 1
# 3 black sofa 1
# 4 red sofa 1
my.summary <- do.call(data.frame, rows.per.group)
colnames(my.summary) <- c(colnames(df)[1:2], 'rows.per.group')
my.data <- merge(df, my.summary, by = c(colnames(df)[1:2]))
my.data
# name type num rows.per.group
# 1 black chair 4 2
# 2 black chair 5 2
# 3 black sofa 12 1
# 4 red plate 3 1
# 5 red sofa 4 1
df这应该可以完成您的工作:
df_agg <- aggregate(num~name+type,df,FUN=NROW)
names(df_agg)[3] <- "count"
df <- merge(df,df_agg,by=c('name','type'),all.x=TRUE)
df_agg一个两行的替代方法是生成一个0的变量,然后用split将行数合并到基本数据集中只差一步
使用broom
软件包中的tidy()
功能,将频率表转换为数据帧,并与df
进行内部连接:
df <- data.frame(name=c('black','black','black','red','red'),
type=c('chair','chair','sofa','sofa','plate'),
num=c(4,5,12,4,3))
library(broom)
df <- merge(df, tidy(table(df[ , c("name","type")])), by=c("name","type"))
df
name type num Freq
1 black chair 4 2
2 black chair 5 2
3 black sofa 12 1
4 red plate 3 1
5 red sofa 4 1
df使用sqldf包:
以R为底的一条简单线:
df$count = table(interaction(df[, (c("name", "type"))]))[interaction(df[, (c("name", "type"))])]
为清晰/高效起见,两行相同:
fact = interaction(df[, (c("name", "type"))])
df$count = table(fact)[fact]
请解释这是如何更一般化的?您是否需要“setkeyv(dt,c('name','type')”?也可以使用transform(df,count=ave(num,name,type,FUN=length))
或with