Probability -> controlled-vocabulary quantile label

`0.025 -> "q02x5"`, `0.5 -> "q50x0"`, `0.975 -> "q97x5"`, `0.005 -> "q00x5"`. Two integer-percent digits, then `x`, then one decimal-percent digit.

Usage

q_label(p)

Arguments

p

Numeric vector of probabilities in [0, 1].

Value

Character vector of quantile labels.

Details

The format is lossy in two ways. It holds one decimal-percent digit, so a finer probability is rounded (`0.0125 -> "q01x2"`). And it holds exactly two integer-percent digits, so `p = 1` produces the three-digit `"q100x0"`, which [q_value] cannot read back. Keep `p` on the 0.001 grid and strictly below 1.

See also

[q_value] reads these labels back, for `p < 1`. vignette("pipeline", package = "csalert") calls this function in its naming-grammar section, and every `_qNNxN` column it prints was labelled by it.

Other naming grammar functions: csfmt_interpret(), csfmt_parse(), csfmt_var(), q_value()

Examples

q_label(c(0.025, 0.5, 0.975))
#> [1] "q02x5" "q50x0" "q97x5"

# limit 1: a probability finer than one decimal percent is rounded
q_value(q_label(0.0125))
#> [1] 0.012

# limit 2: p = 1 produces a three-digit label q_value() returns NA for
q_label(1)
#> [1] "q100x0"
q_value(q_label(1))
#> [1] NA