How to fill in the preceding numbers whenever there is a 0 in R?

Question

I have a string of numbers:

n1 = c(1, 1, 0, 6, 0, 0, 10, 10, 11, 12, 0, 0, 19, 23, 0, 0)

I need to replace 0 with the corresponding number righ

Answer 1

Try na.locf() from the package zoo:

library(zoo)
n1 <- c(1, 1, 0, 6, 0, 0, 10, 10, 11, 12, 0, 0, 19, 23, 0, 0)
n1[n1 == 0] <- NA
na.locf(n1)
## [1]  1  1  1  6  6  6 10 10 11 12 12 12 19 23 23 23

This function replaces each NA with the most recent non-NA prior to it. This is why I substituted all 0s with NA before applying it.

Here's a discussion on a similar (yet not identical) issue.

EDIT: If n1 eventually consists of NAs, try e.g.

n1 <- c(1, 1, 0, 6, 0, 0, 10, NA, 11, 12, 0, 0, 19, NA, 0, 0)
wh_na <- which(is.na(n1))
n1[n1 == 0] <- NA
n2 <- na.locf(n1)
n2[wh_na] <- NA
n2
##  [1]  1  1  1  6  6  6 10 NA 11 12 12 12 19 NA 19 19

EDIT2: This approach for c(1,NA,0) returns c(1,NA,1). The other two funs give c(1,NA,NA). In other words, here we're replacing 0 with last non-missing, non-zero value. Choose your favourite option.

EDIT3: Inspired by @Thell's Rcpp solution, I'd like to add another one - this time using "pure" R/C API.

library('inline')
sexp0 <- cfunction(signature(x="numeric"), "
   x = Rf_coerceVector(x, INTSXP); // will not work for factors
   R_len_t n = LENGTH(x);
   SEXP ret;
   PROTECT(ret = Rf_allocVector(INTSXP, n));
   int lval = NA_INTEGER;
   int* xin = INTEGER(x);
   int* rin = INTEGER(ret);
   for (R_len_t i=0; i

In this case we will get c(1,NA,NA) for c(1,NA,0). Some benchmarks:

library(microbenchmark)
set.seed(1L)
n1 <- sample(c(0:10), 1e6, TRUE)
microbenchmark(sexp0(n1), rollValue(n1), n1[cummax(seq_along(n1) * (n1 != 0))])
## Unit: milliseconds
##                                   expr       min        lq    median        uq       max neval
##                              sexp0(n1)  2.468588  2.494233  3.198711  4.216908  63.21236   100
##                          rollValue(n1)  8.151000  9.359731 10.603078 12.760594  75.88901   100
##  n1[cummax(seq_along(n1) * (n1 != 0))] 32.899420 36.956711 39.673726 45.419449 106.48180   100