ทำความเข้าใจ NA ใน R

# NA represents a missing value
height_cm <- c(175, NA, 162, 188, NA, 170)
cat('Heights:', height_cm, '
')
cat('Length:', length(height_cm), '
')  # NA counts as an element
cat('Sum:', sum(height_cm), '
')        # NA propagates

# NA: element exists, value unknown
vec_with_na <- c(1, NA, 3)
cat('Vector with NA, length:', length(vec_with_na), '
')  # 3

# NULL: nothing there at all
vec_with_null <- c(1, NULL, 3)
cat('Vector with NULL, length:', length(vec_with_null), '
')  # 2!

# NULL gets dropped; NA is kept
cat('vec_with_na:', vec_with_na, '
')
cat('vec_with_null:', vec_with_null, '
')

# NaN from undefined operations
cat('0/0 =', 0/0, '
')          # NaN
cat('sqrt(-1) =', sqrt(-1), '
') # NaN (with warning)
cat('Inf - Inf =', Inf - Inf, '
') # NaN

# NaN IS also NA (a special case)
cat('is.nan(NaN):', is.nan(NaN), '
')
cat('is.na(NaN): ', is.na(NaN), '
')  # TRUE!
cat('is.nan(NA): ', is.nan(NA), '
')  # FALSE

cat('1/0 =', 1/0, '
')         # Inf
cat('-1/0 =', -1/0, '
')       # -Inf
cat('Inf + 1 =', Inf + 1, '
') # Inf
cat('Inf > 1000:', Inf > 1000, '
')  # TRUE
cat('is.infinite(Inf):', is.infinite(Inf), '
')
cat('is.na(Inf):       ', is.na(Inf), '
')  # FALSE

x <- NA
cat('NA + 5:  ', x + 5, '
')      # NA
cat('NA * 0:  ', x * 0, '
')      # NA (not 0!)
cat('NA == NA:', x == NA, '
')     # NA (not TRUE!)
cat('NA > 5:  ', x > 5, '
')      # NA
cat('sum(1,NA):', sum(1, NA), '
') # NA

cat('typeof(NA):            ', typeof(NA), '
')           # logical
cat('typeof(NA_integer_):  ', typeof(NA_integer_), '
')   # integer
cat('typeof(NA_real_):     ', typeof(NA_real_), '
')      # double
cat('typeof(NA_complex_):  ', typeof(NA_complex_), '
')   # complex
cat('typeof(NA_character_):', typeof(NA_character_), '
') # character

# Using plain NA in an integer vector
v1 <- c(1L, 2L, NA, 4L)
cat('typeof with NA:         ', typeof(v1), '
')  # integer (OK here)

# Explicit typed NA is safer in functions
v2 <- c(1L, 2L, NA_integer_, 4L)
cat('typeof with NA_integer_:', typeof(v2), '
')  # integer

# Check they behave the same
cat('Same?', identical(v1, v2), '
')

# Character vector with missing
names_vec <- c('Alice', 'Bob', NA_character_, 'Dave')
cat('Names:', names_vec, '
')
cat('typeof:', typeof(names_vec), '
')

# Numeric with missing
readings <- c(22.5, NA_real_, 21.3, NA_real_, 24.1)
cat('Readings:', readings, '
')
cat('Non-NA count:', sum(!is.na(readings)), '
')

cat('TRUE  | NA:', TRUE | NA, '
')   # TRUE
cat('FALSE | NA:', FALSE | NA, '
')  # NA
cat('TRUE  & NA:', TRUE & NA, '
')   # NA
cat('FALSE & NA:', FALSE & NA, '
')  # FALSE

# Never use == to test for NA!
cat('NA == NA:', NA == NA, '
')    # NA (not TRUE!)
cat('is.na(NA):', is.na(NA), '
')  # TRUE (correct way)

ages <- c(25, NA, 32, NA, 29, 41, NA, 35)
cat('Ages:', ages, '
')
cat('Length:', length(ages), '
')
cat('Missing count:', sum(is.na(ages)), '
')
cat('Present count:', sum(!is.na(ages)), '
')
cat('Mean (fails):', mean(ages), '
')           # NA
cat('Mean (works):', mean(ages, na.rm = TRUE), '
') # 32.4

# Test functions for special values
x <- c(1, NA, NaN, Inf, -Inf, 0)
cat('Values:         ', x, '
')
cat('is.na():        ', is.na(x), '
')        # TRUE for NA and NaN
cat('is.nan():       ', is.nan(x), '
')       # TRUE only for NaN
cat('is.infinite():  ', is.infinite(x), '
')  # TRUE for Inf and -Inf
cat('is.finite():    ', is.finite(x), '
')    # TRUE only for normal numbers

# Key tests side by side
special_vals <- list(NA=NA, NaN=NaN, Inf=Inf, 'NULL-in-vec'=c(1,NULL,3)[2])
for (name in names(special_vals)) {
  v <- special_vals[[name]]
  cat(name, '| is.na:', is.na(v), '| is.nan:', is.nan(v[!is.infinite(v)]), '
')
}