使用R按因子级别查找重叠范围
我需要在单个数据集中找到重叠的范围,但需要为每个ID或因子级别找到它们。任何帮助都将不胜感激使用R按因子级别查找重叠范围,r,R,我需要在单个数据集中找到重叠的范围,但需要为每个ID或因子级别找到它们。任何帮助都将不胜感激 library(dplyr) df_foo = read.table( textConnection("Class Min Max A 500 630 A 100 200 B 100 200 A 210 310 A 200 210 B 210 310 A 510 530 B 200 210 A 705 800
library(dplyr)
df_foo = read.table(
textConnection("Class Min Max
A 500 630
A 100 200
B 100 200
A 210 310
A 200 210
B 210 310
A 510 530
B 200 210
A 705 800
B 500 630
B 510 530
B 705 800"), header = TRUE
)
c = outer(df_foo$Max, df_foo$Min, ">")
d = outer(df_foo$Min, df_foo$Max, "<")
df_foo %>%
mutate(Overlap = apply(c & d, 1, sum) > 1
)
但我想找出A和B的每个级别之间的重叠,如下所示:
Class Min Max Overlap
1 A 500 630 TRUE
2 A 100 200 FALSE
3 B 100 200 FALSE
4 A 210 310 FALSE
5 A 200 210 FALSE
6 B 210 310 FALSE
7 A 510 530 TRUE
8 B 200 210 FALSE
9 A 705 800 FALSE
10 B 500 630 TRUE
11 B 510 530 TRUE
12 B 705 800 FALSE
使用
dplyr
df=df_foo%>%group_by(Class)%>%
mutate(Overlap=if_else(Min<lag(Max,order_by=Class),TRUE,FALSE))
df$Overlap[which(df$Overlap==TRUE)-1]=TRUE
df$Overlap[which(is.na(df$Overlap))]=FALSE
> df
# A tibble: 12 x 4
# Groups: Class [2]
Class Min Max Overlap
<fct> <dbl> <dbl> <lgl>
1 A 100 200 FALSE
2 A 200 210 FALSE
3 A 210 310 FALSE
4 A 500 630 TRUE
5 A 510 530 TRUE
6 A 705 800 FALSE
7 B 100 200 FALSE
8 B 200 210 FALSE
9 B 210 310 FALSE
10 B 500 630 TRUE
11 B 510 530 TRUE
12 B 705 800 FALSE
df=df\u foo%%>%group\u by(Class)%%>%
变异(重叠=if_else(最小df
#一个tibble:12x4
#组别:班级[2]
类最小最大重叠
1 A 100 200错误
2 A 200 210错误
3 A 210 310错误
4 A 500 630正确
5 A 510 530正确
6 A 705 800假
7 B 100 200错误
8B200210错误
9 B 210 310假
10 B 500 630正确
11 B 510 530正确
12 B 705 800错误
此代码假定您的值按升序排列,因为它只检查前一行
编辑不是最漂亮的,而是有效的
df_foo$Class=as.character.factor(df_foo$Class)
df_foo=as.data.frame(df_foo)
df_foo$Overlap=rep("FALSE",nrow(df_foo))
for (i in 1:nrow(df_foo)){
aux=FALSE
class=df_foo$Class[i]
df=df_foo[-i,]%>%filter(.,Class==class)
for (j in 1:nrow(df)){
if (df_foo[i,"Min"]<df[j,"Max"] & df_foo[i,"Max"] > df[j,"Min"]){
aux=TRUE
}
}
df_foo[i,"Overlap"]=aux
}
> df_foo
Class Min Max Overlap
1 A 500 630 TRUE
2 A 100 200 FALSE
3 B 100 200 FALSE
4 A 210 310 FALSE
5 A 200 210 FALSE
6 B 210 310 FALSE
7 A 510 530 TRUE
8 B 200 210 FALSE
9 A 705 800 FALSE
10 B 500 630 TRUE
11 B 510 530 TRUE
12 B 705 800 FALSE
df_foo$Class=as.character.factor(df_foo$Class)
df_foo=as.data.frame(df_foo)
df_foo$Overlap=rep(“假”,nrow(df_foo))
对于(i in 1:nrow(df_foo)){
aux=错误
class=df_foo$class[i]
df=df_foo[-i,]%>%filter(,Class==Class)
对于(1中的j:nrow(df)){
if(df_foo[i,“Min”]df[j,“Min”]){
aux=真
}
}
df_foo[i,“重叠”]=aux
}
>德福
类最小最大重叠
1 A 500 630正确
2 A 100 200错误
3B100 200错误
4 A 210 310错误
5 A 200 210错误
6B210310错误
7 A 510 530正确
8B200210错误
9 A 705 800假
10 B 500 630正确
11 B 510 530正确
12 B 705 800错误
必须有一种方法可以使用
dplyr
来实现,但我无法找到它。所发生的事情是,它循环通过df_foo
的每一行;它生成一个dataframe
与同一组的所有其他行,并比较是否有重叠(min我在数据中得到了答案。表
,翻译成dplyr
应该是straigtworfard。其想法是为每个类别创建一个先前累积最大值的向量:
df_foo <- setDT(df_foo)
df_foo[, shiftedmaxmax := c(NA,cummax(Max)[1:(.N-1)]),by = Class ]
Class Min Max shiftedmaxmax
1: A 100 200 NA
2: A 200 210 200
3: A 210 310 210
4: A 500 630 310
5: A 510 530 630
6: A 705 800 630
7: B 100 200 NA
8: B 200 210 200
9: B 210 310 210
10: B 500 630 310
11: B 510 530 630
12: B 705 800 630
另一种数据表方法。
在这个答案中,样本数据/范围的顺序是不相关的…foverlaps()
为您做了所有艰苦的工作
样本数据
library( data.table )
dt <- as.data.table( df_foo )
库(data.table)
dt谢谢,我本应该更清楚,但在我的实际数据集中,所有值都是混合的,不遵循升序或降序。是否有多个重叠?如果没有,只需按类
按升序重新排列,代码就可以了。如果没有,应该清楚地更改。可能有多个重叠p、 我编辑了该示例以显示数据是如何混淆的。感谢您的帮助!谢谢,不幸的是,范围不是按升序或降序排列的。它们都混淆了。我编辑了数据集以反映这一点。
df_foo[,superposed := Min < shiftedmaxmax]
Class Min Max shiftedmaxmax superposed
1: A 100 200 NA NA
2: A 200 210 200 FALSE
3: A 210 310 210 FALSE
4: A 500 630 310 FALSE
5: A 510 530 630 TRUE
6: A 705 800 630 FALSE
7: B 100 200 NA NA
8: B 200 210 200 FALSE
9: B 210 310 210 FALSE
10: B 500 630 310 FALSE
11: B 510 530 630 TRUE
12: B 705 800 630 FALSE
df_foo[,superposedsource := Max %in% shiftedmaxmax[superposed],by = Class]
df_foo[,superposedtot := ifelse((superposed | superposedsource) &,T,F)]
Class Min Max shiftedmaxmax superposed superposedsource superposedtot
1: A 100 200 NA NA FALSE NA
2: A 200 210 200 FALSE FALSE FALSE
3: A 210 310 210 FALSE FALSE FALSE
4: A 500 630 310 FALSE TRUE TRUE
5: A 510 530 630 TRUE FALSE TRUE
6: A 705 800 630 FALSE FALSE FALSE
7: B 100 200 NA NA FALSE NA
8: B 200 210 200 FALSE FALSE FALSE
9: B 210 310 210 FALSE FALSE FALSE
10: B 500 630 310 FALSE TRUE TRUE
11: B 510 530 630 TRUE FALSE TRUE
12: B 705 800 630 FALSE FALSE FALSE
library( data.table )
dt <- as.data.table( df_foo )
#set key for the data.table
setkey(dt, Min, Max)
#perform overlap join, keep only joined ranges where the class is the same, and Min and Max are not the same.
result <- foverlaps( dt, dt )[ Class == i.Class & !(Min == i.Min | Max == i.Max | Min == i.Max | Max == i.Min), ]
#create a logical vector (i.e. Overlap) by checking if the (pasted) combination of
#Class, Min and Max exists in both 'dt' and 'result'
dt[ , Overlap := paste0( Class, Min, Max ) %in% paste0( result$Class, result$Min, result$Max) ][]
# Class Min Max Overlap
# 1: A 100 200 FALSE
# 2: B 100 200 FALSE
# 3: A 200 210 FALSE
# 4: B 200 210 FALSE
# 5: A 210 310 FALSE
# 6: B 210 310 FALSE
# 7: A 500 630 TRUE
# 8: B 500 630 TRUE
# 9: A 510 530 TRUE
# 10: B 510 530 TRUE
# 11: A 705 800 FALSE
# 12: B 705 800 FALSE