Question

我正在尝试创建一个函数，该函数将为数据集中的一组选定列创建新的标记指示符列。

# Data Set
A = as.factor(c(0,2,1,0))
B = as.factor(c(2,NA,1,0))
C = as.factor(c(1,0,NA,0))
D = as.factor(c(NA,2,0,1))
dat = data.table(A, B, C, D)

目前我正在为我想要的每一栏做这件事：

# What I'm currently doing (expected output of loop matches these columns)
attach(dat)
        VAR = B
        dat$b.test[VAR == "0"] <- "0"
        dat$b.test[VAR == "1" | VAR == "2"] <- "1"

        VAR = C
        dat$c.test[VAR == "0"] <- "0"
        dat$c.test[VAR == "1" | VAR == "2"] <- "1"

        VAR = D
        dat$d.test[VAR == "0"] <- "0"
        dat$d.test[VAR == "1" | VAR == "2"] <- "1"
detach(dat)

似乎我应该能够创建一个for循环，它将运行一个包含我希望逻辑在（B，C，D）上执行的所有列的向量，并从另一个向量中调用一个新名称（b2），c2，d2）在每个向量内的相同位置。

尝试方法1

# Failed method 1
attach(dat)
        new.var = c(b2, c2, d2)
        cur.var = c(B, C, D)
        l = length(cur.var)

        for(i in 1:l){
                X = cur.var[i]
                VAR = cur.var[i]
                dat$X[VAR == "0"] <- "0"
                dat$X[VAR == "1" | VAR == "2"] <- "1"
        }
detach(dat)

新专栏X的结果

尝试方法2

# Failed method 2
        new.var = c(dat$b2, dat$c2, dat$d2)
        cur.var = c(dat$B, dat$C, dat$D)
        l = length(cur.var)
for(i in 1:l){
        new.var[i] = ifelse(new.var[i] == "0", "0",
                            ifelse(new.var[i] == "1" | "2", "1", NA)
                            )
}

是否有不同的方法可以尝试这样做？

Answer 1

你真的不需要循环。您可以使用.SD变量迭代列。例如

change<-c("B","C","D")

myfun <- function(x) ifelse(x==0, "0", "1")
dat[,paste0(change,".test") := Map(myfun, .SD), .SDcols=change]

#    A  B  C  D B.test C.test D.test
# 1: 0  2  1 NA      1      1     NA
# 2: 2 NA  0  2     NA      0      1
# 3: 1  1 NA  0      1     NA      0
# 4: 0  0  0  1      0      0      1

创建循环通过2个向量的列

1 个答案: