R/confusion_matrix-package.R

#' @name confusionMatrix

#' @description  The goal of this package is primarily to provide an easy way to
#'   obtain common confusion table metrics in a tidy fashion.  The inspiration
#'   comes from Max Kuhn's \code{caret} package and associated function
#'   \code{confusionMatrix}, and the continuation of those efforts in the
#'   \code{yardstick} package.  Here, practically all dependencies have been
#'   removed except for dplyr, and results are tibbles making for easier
#'   document presentation, as well as the ability to peel off the statistics
#'   desired.
#'
#'   All that is required is a vector of predicted classes and a vector of
#'   target classes, as that is typically what we're dealing with in such
#'   scenarios, i.e. predictions vs. a target variable.  These can be logical,
#'   integer/numeric, character, or factor, but the predictions should match the
#'   target in an obvious way.
#'
#'   Statistics provided include:
#'
#'   Accuracy and Agreement
#'
#'     - Accuracy, bounds, and related
#'
#'     - Cohen's Kappa
#'
#'     - Corrected Rand
#'
#'   Other Statistics:
#'
#'     - Sensitivity
#'
#'     - Specificity
#'
#'     - Prevalence
#'
#'     - Positive Predictive Value
#'
#'     - Negative Predictive Value
#'
#'     - Detection prevalence
#'
#'     - Balanced Accuracy
#'
#'     - F1
#'
#'   Measures of Agreement/Association:
#'
#'     - Phi
#'
#'     - Yule's
#'
#'     - Peirce's science of the method (Youden's J)
#'
#'     - Jaccard
#'

#' @keywords internal
"_PACKAGE"

# The following block is used by usethis to automatically manage
# roxygen namespace tags. Modify with care!
## usethis namespace: start
## usethis namespace: end

NULL
m-clark/confusionMatrix documentation built on July 15, 2020, 4:16 p.m.