※ 引述《Edster (Edster)》之銘言：
: best <- function(x, y){
: z <- c("heart attack", "heart failure", "pneumonia")
: outcome <- read.csv(file="outcome-of-care-measures.csv", header=T, as.is=T)
: SS <- as.numeric(outcome[,11])
: if (! (x %in% outcome$State) | ! (y %in% z)){
: stop("Error in best(state, outcome) : invalid state")
: }
: if (y == "heart attack"){
: select = outcome$State == x & !is.na(SS); o = order(SS)
: result <- outcome[o[select],c(1,2)]
: print(paste("best (", result, ") heart attack"))
: }
: }
請問從倒數第三行(select = outcome$State....)之後，
為何我用
o <- order(SS[select])
outcome.order <- outcome[o, ]
結果outcome.order還是有NA值存在？

看來問題出現在order. 那麼就得要老老實實的把order 跟 is.na 分開了.outcome = outcome[order(as.numeric(outcome[,11])),]select = outcome$State == x & !is.na((as.numeric(outcome[,11]))result <- outcome[select,c(1,2)]變得有點醜... 現下也想不到更好的方式.