confusionMatrix: confusionMatrix: Confusion Matrix
In m-clark/confusionMatrix: Confusion Matrix

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 caret package and associated function confusionMatrix, and the continuation of those efforts in the 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