performance_classification()

We should add a function that computes typical classification indices derived from the confusion matrix. We could perhaps, as a start, wrap around the caret function, even though I think that the underlying maths are fairly simple.

pred <-kmeans(iris[1:4], 3)$cluster

caret::confusionMatrix(as.factor(pred), as.factor(as.numeric(iris$Species)))
#> Confusion Matrix and Statistics
#> 
#>           Reference
#> Prediction  1  2  3
#>          1 50  0  0
#>          2  0 48 14
#>          3  0  2 36
#> 
#> Overall Statistics
#>                                           
#>                Accuracy : 0.8933          
#>                  95% CI : (0.8326, 0.9378)
#>     No Information Rate : 0.3333          
#>     P-Value [Acc > NIR] : < 2.2e-16       
#>                                           
#>                   Kappa : 0.84            
#>                                           
#>  Mcnemar's Test P-Value : NA              
#> 
#> Statistics by Class:
#> 
#>                      Class: 1 Class: 2 Class: 3
#> Sensitivity            1.0000   0.9600   0.7200
#> Specificity            1.0000   0.8600   0.9800
#> Pos Pred Value         1.0000   0.7742   0.9474
#> Neg Pred Value         1.0000   0.9773   0.8750
#> Prevalence             0.3333   0.3333   0.3333
#> Detection Rate         0.3333   0.3200   0.2400
#> Detection Prevalence   0.3333   0.4133   0.2533
#> Balanced Accuracy      1.0000   0.9100   0.8500

^{Created on 2021-08-11 by the reprex package (v2.0.1)}

One of the things that we could improve is the need to relevel / modify the input so that is has exactly the same levels and level types.

The issue names no implementation file, test, or entry point. Start by reviewing the caret::confusionMatrix example and the project's existing performance metrics; done requires an agreed set of classification indices and consistent handling of prediction and reference levels.