Replace -inf, NaN and NA values with zero in a dataset in R
Andrew Mclaughlin
I am trying to run some trading strategies in R. I have downloaded some stock prices and calculated returns. The new return dataset has a number of -inf, NaN, and NA values. I am reproducing a row of the dataset (log_ret). Its a zoo dataset.
library(zoo)
log_ret <- structure( c(0.234,-0.012,-Inf,NaN,0.454,Inf), .Dim = c(1L, 6L), .Dimnames = list(NULL, c("x", "y", "z", "s", "p", "t")), index = structure(12784, class = "Date"), class = "zoo"
) x y z s p t
2005-01-01 0.234 -0.012 -Inf NaN 0.454 InfHow can I replace these unwanted values with 0?
15 Answers
As per ?zoo:
Subscripting by a zoo object whose data contains logical values is undefined.
So you need to wrap the subsetting in a which call:
log_ret[which(!is.finite(log_ret))] <- 0
log_ret x y z s p t
2005-01-01 0.234 -0.012 0 0 0.454 0 0 Inf, NA and NaN are matched by !is.finite, for example
a <- c(1, Inf, NA, NaN)
a[!is.finite(a)] <- 0
# a is now [1, 0, 0, 0]I don't know too much about manipulating zoo objects, but for the example above
log_ret[1, !is.finite(log_ret)] <- 0works. In your actual data you will have to loop over all rows. There might be a zoo-specific way of doing this.
Edit: The zoo-specific way is log_ret[which(!is.finite(log_ret))] <- 0.
Another way to do it is (where df=your dataframe):
is.na(df)<-sapply(df, is.infinite)
df[is.na(df)]<-0I don't know if this works for zoo objects, but it gets around the problem of is.infinite() only working on vectors.
1Using mutate_all in dplyr:
library(dplyr)
fortify.zoo(log_ret) %>% mutate_all(function(x) ifelse(is.infinite(x), 0, x)) 2 Since the lifecycle for mutate_all has been superseded by the use of across:
library(dplyr)
fortify.zoo(log_ret) %>% mutate(across(.cols = everything(), ~ ifelse(is.infinite(.x), 0, .x)))