R ignoring NA value?

Sometimes it might be very confusing when it looks like R is ignoring the NA value. When some of the code relying on that values, the results might be confusing.

The problem, when it looks like R ignores the NA value, appears when some of the NA values are a string. It might happen after using some of the functions. If the column already contains NA, then sprintf might turn it into a string format.

Here is my dataset that contains column values with missing values.

df <- data.frame(
  agent = as.character(c("David", "David", "David", "Kate", "Kate", "Alma", "Alma")),
  value = as.numeric(c(701.8299, 698.7306, NA, NA, 698.7125, 698.5938, 690.8382))
)

df
#  agent    value
#1 David 701.8299
#2 David 698.7306
#3 David       NA
#4  Kate       NA
#5  Kate 698.7125
#6  Alma 698.5938
#7  Alma 690.8382

Here is an additional step with ifelse checking some logic. If it is not true, ifelse returns the column with improvements from sprintf.

df$value <- ifelse(df$agent == "Kate", NA, sprintf("%.3f", df$value))

As you can see, that visually there is some difference between NA values. There is no surprise that R ignoring NA value that only looks like NA.

is.na(df$value)
FALSE FALSE FALSE  TRUE  TRUE FALSE FALSE

The fastest way to convert string NA values is with the function as.numeric.

df$value <- as.numeric(df$value)

NA values may cause some big headaches. Take a look at one of the examples with ifelse: ifelse and NA problem in R.