如何将所有非数字值转换为数据帧中R中的NA？

Question

I have a dataframe called returns. How do I make the code below more concise? I tried using apply, lappply and sapply, but they seem to change the structure of the dataframe such that I cannot produce a correlation matrix. To clarify: the code below works, but I would like to make it more concise. Thanks!

returns$VIT <- as.numeric(as.character(returns$VIT))
returns$EFA <- as.numeric(as.character(returns$EFA))
returns$VWO <- as.numeric(as.character(returns$VWO))
returns$VIG <- as.numeric(as.character(returns$VIG))
returns$VNQ <- as.numeric(as.character(returns$VNQ))
returns$iPath.DJP <- as.numeric(as.character(returns$iPath.DJP))
returns$iShares.MUB <- as.numeric(as.character(returns$iShares.MUB))

cor(returns, use="pairwise.complete.obs")

EDIT: The following test code does NOT work, I am trying to figure out how to make it work.

test <- data.frame(c(.04,.2,"blah"),c(.01,.24,"blah"))
colnames(test) <- c("VIT", "EFA")

new <- apply(test, 2, function(x) as.numeric(as.character(x)))
cor(test, use="pairwise.complete.obs")

Answer 1

test <- data.frame(a = rep("bob", 12), b = rep(c(1,23,4), times = 4))
Filter(is.numeric, test)

Returns only the numeric columns in test . You can use lapply(test, is.numeric) to get the true/false results yourself and do something with that.

Answer 2

Is this what you're looking for?

numeric.columns <- c('VIT','EFA','VWO','VIG','VNQ','iPath.DJP','iPath.DJP','iShares.MUB')
returns[,numeric.columns] <- lapply(returns[,numeric.columns], function(x) as.numeric(as.character(x)))
cor(returns, use="pairwise.complete.obs")

Answer 3

Just use plyr 's colwise .

returns = colwise(function(x) as.numeric(ifelse(is.numeric(x),x,NA)))(returns)

It takes a function and applies it to all columns. The final result should be numeric. Note that this treats values such as '1' as NA.