选择连续值更改而忽略NA的行

Question

I want to find rows where value changes within a vector that contains NA s. For example, I have a following vector

x = c(1, 2, 6, 3, 3, 9, 3, 5, 5, 4, 4)

I can use this code which(x[-1] != x[-length(x)]) + 1 and it will give me row numbers when values where changed: 2, 3, 4, 6, 7, 8, 10 .

If I replace x[x>4] <- NA and rerun the same code, the output will be the row 2 only. But I want to consider all NA s as changes except if the previous value is also NA and deal with the value after as change.

My desired output for x = c(1, 2, NA, 3, 3, NA, 3, NA, NA, 4, 4) would be output of following row numbers: 2, 3, 4, 6, 7, 8, 10

Answer 1

I would recommend to change NA with other conventional representation of missing values, such as -9999. After this you can use your method which(x[-1] != x[-length(x)]) + 1 , or try rle function from base R.

# Sample data
x = c(1, 2, NA, 3, 3, NA, 3, NA, NA, 4, 4)

# Replace missing values with -9999
x[is.na(x)] <- -9999

# Calculate position of non-equal consecutive values
cumsum(rle(x)$length) + 1

# NOTE: you will need to remove last element of the output