R/class-detection_prevalence.R

In yardstick: Tidy Characterizations of Model Performance

#' Detection prevalence
#'
#' Detection prevalence is defined as the number of _predicted_ positive events (both
#' true positive and false positive) divided by the total number of predictions.
#'
#' @family class metrics
#' @templateVar fn detection_prevalence
#' @template event_first
#' @template multiclass
#' @template return
#'
#' @inheritParams sens
#'
#' @author Max Kuhn
#'
#' @template examples-class
#'
#' @export
detection_prevalence <- function(data, ...) {
  UseMethod("detection_prevalence")
}
detection_prevalence <- new_class_metric(
  detection_prevalence,
  direction = "maximize"
)

#' @export
#' @rdname detection_prevalence
detection_prevalence.data.frame <- function(data,
                                            truth,
                                            estimate,
                                            estimator = NULL,
                                            na_rm = TRUE,
                                            case_weights = NULL,
                                            event_level = yardstick_event_level(),
                                            ...) {
  class_metric_summarizer(
    name = "detection_prevalence",
    fn = detection_prevalence_vec,
    data = data,
    truth = !!enquo(truth),
    estimate = !!enquo(estimate),
    estimator = estimator,
    event_level = event_level,
    na_rm = na_rm,
    case_weights = !!enquo(case_weights)
  )
}

#' @export
detection_prevalence.table <- function(data,
                                       estimator = NULL,
                                       event_level = yardstick_event_level(),
                                       ...) {
  check_table(data)
  estimator <- finalize_estimator(data, estimator)

  metric_tibbler(
    .metric = "detection_prevalence",
    .estimator = estimator,
    .estimate = detection_prevalence_table_impl(data, estimator, event_level)
  )
}

#' @export
detection_prevalence.matrix <- function(data,
                                        estimator = NULL,
                                        event_level = yardstick_event_level(),
                                        ...) {
  data <- as.table(data)
  detection_prevalence.table(data, estimator, event_level)
}

#' @export
#' @rdname detection_prevalence
detection_prevalence_vec <- function(truth,
                                     estimate,
                                     estimator = NULL,
                                     na_rm = TRUE,
                                     case_weights = NULL,
                                     event_level = yardstick_event_level(),
                                     ...) {
  abort_if_class_pred(truth)
  estimate <- as_factor_from_class_pred(estimate)

  estimator <- finalize_estimator(truth, estimator)

  check_class_metric(truth, estimate, case_weights, estimator)

  if (na_rm) {
    result <- yardstick_remove_missing(truth, estimate, case_weights)

    truth <- result$truth
    estimate <- result$estimate
    case_weights <- result$case_weights
  } else if (yardstick_any_missing(truth, estimate, case_weights)) {
    return(NA_real_)
  }

  data <- yardstick_table(truth, estimate, case_weights = case_weights)
  detection_prevalence_table_impl(data, estimator, event_level)
}

detection_prevalence_table_impl <- function(data, estimator, event_level) {
  if (is_binary(estimator)) {
    detection_prevalence_binary(data, event_level)
  } else {
    w <- get_weights(data, estimator)
    out_vec <- detection_prevalence_multiclass(data, estimator)
    stats::weighted.mean(out_vec, w)
  }
}

detection_prevalence_binary <- function(data, event_level) {
  pos_level <- pos_val(data, event_level)
  sum(data[pos_level, ]) / sum(data)
}

detection_prevalence_multiclass <- function(data, estimator) {
  numer <- rowSums(data)
  denom <- rep(sum(data), times = nrow(data))

  denom[denom <= 0] <- NA_real_

  if (is_micro(estimator)) {
    numer <- sum(numer)
    denom <- sum(denom)
  }

  numer / denom
}