lag function in ifelse - r

Question

I have a dataframe df with a structure like this:
val1 val2 val3
1 12 NA
2 14 NA
3 54 54
1 35 4
2 3 5
3 7 NA
4 8 NA
5 9 NA
Expected value:
val1 val2 val3 val4
1 12 NA 12
2 14 NA 12
3 54 54 54
1 35 4 35
2 3 5 3
3 7 NA 3
4 8 NA 3
5 9 NA 3
Problem:
I need a new column val4 with the following condition
df$val4 <- ifelse(df$val1 == 1, df$val2, ifelse(is.na(df$val3), lag(df$val4), df$val2))
This leads to
Error in hasTsp(x) : attempt to set an attribute on NULL
Condition:
val4 is equal to value of val2 when val1 is equal 1 (val3 does not matter)
val4 is equal to previous value when val3 is NA ( expect when val1 is not equal to 1)
P.S: I know I can use for loop here, but that would be very slow!

Answer 1

We can use data.table with zoo. Convert the 'data.frame' to 'data.table' (setDT(df)), create the 'val4' by multiplying the 'val2' with a vector of 1's and NA (NA^is.na(val3) - returns NA for NA elements in 'val3' while the non-NA is changed to 1), then for 1 in 'val1', we assign 'val4' to 'val2', and replace the NA elements with the previous non-NA elements with na.locf
library(data.table)
library(zoo)
setDT(df)[, val4 := val2 * NA^is.na(val3)
][val1==1, val4 := val2
][, val4 := na.locf(val4)][]
# val1 val2 val3 val4
#1: 1 12 NA 12
#2: 2 14 NA 12
#3: 3 54 54 54
#4: 1 35 4 35
#5: 2 3 5 3
#6: 3 7 NA 3
#7: 4 8 NA 3
#8: 5 9 NA 3
More code explanation
`is.na` returns a `logical` vector
setDT(df)[, is.na(val3)]
#[1] TRUE TRUE FALSE FALSE FALSE TRUE TRUE TRUE
If we need to change the TRUE values to NA and 1 to others
setDT(df)[, NA^is.na(val3)]
#[1] NA NA 1 1 1 NA NA NA
Multiply by 'val2'
setDT(df)[, val2 * NA^is.na(val3)]
#[1] NA NA 54 35 3 NA NA NA
and the rest is just assignment based on the logical condition in 'i'
data
df <- structure(list(val1 = c(1L, 2L, 3L, 1L, 2L, 3L, 4L, 5L), val2 = c(12L,
14L, 54L, 35L, 3L, 7L, 8L, 9L), val3 = c(NA, NA, 54L, 4L, 5L,
NA, NA, NA)), .Names = c("val1", "val2", "val3"), class = "data.frame",
row.names = c(NA, -8L))