[英]Row mean of selected columns conditional on a different column
假設在data.table中有許多模擬(和其他變量):
data <- setDT(data.frame(sim1=c(1,1,1), sim2= c(2,2,2), sim3=c(3,3,3),
sim4=c(4,4,4), sim5=c(5,5,5), index=c(2,2,2)))
sim1 sim2 sim3 sim4 sim5 index
1: 1 2 3 4 5 2
2: 1 2 3 4 5 2
3: 1 2 3 4 5 2
我要計算高於索引列的模擬平均值:
data[, higher.than.index.ave := rowMeans(.SD[.SD > index]),
.SDcols = names(data[, grepl(paste(paste("sim", 1:5, sep=""),
collapse = "|") , names(data)), with=FALSE])]
我也嘗試過其他解決方案,但是沒有運氣。 有什么建議可以執行這樣的任務嗎?
data <- data.table(sim1=c(1,1,1), sim2= c(2,2,2), sim3=c(3,3,3),
sim4=c(4,4,4), sim5=c(5,5,5), index=c(2,2,2))
data[, means :=
rowMeans(data[, lapply(.SD, function(x) ifelse(x < index, NA, x))
][, -'index'],
na.rm = T)]
或者,使用.SDcols
僅選擇sim
列:
data[, means :=
rowMeans(data[, lapply(.SD, function(x) ifelse(x < index, NA, x))
, .SDcols = intersect(paste0('sim', 1:5), names(data))],
na.rm = T)]
輸出:
data
sim1 sim2 sim3 sim4 sim5 index means
1: 1 2 3 4 5 2 3.5
2: 1 2 3 4 5 2 3.5
3: 1 2 3 4 5 2 3.5
data$higher.than.index.ave <- apply(data,1,function(x) {y <- x[1:5]; mean(y[y>=x[6]])})
# sim1 sim2 sim3 sim4 sim5 index higher.than.index.ave
# 1: 1 2 3 4 5 2 3.5
# 2: 1 2 3 4 5 2 3.5
# 3: 1 2 3 4 5 2 3.5
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.