Closed lshpaner closed 1 month ago
Description:
The confusion matrix printout has misaligned "Pos" and "Neg" labels under the "Predicted" and "Actual" sections. This leads to incorrect associations between the labels and the corresponding confusion matrix values.
The return_metrics generates the following confusion matrix using the AIDS Clinical Trials Notebook as an example:
return_metrics
Confusion matrix on set provided: -------------------------------------------------------------------------------- Predicted: Pos Neg -------------------------------------------------------------------------------- Actual: Pos 304 (tp) 20 (fn) Neg 40 (fp) 64 (tn) -------------------------------------------------------------------------------- precision recall f1-score support 0 0.88 0.94 0.91 324 1 0.76 0.62 0.68 104 accuracy 0.86 428 macro avg 0.82 0.78 0.80 428 weighted avg 0.85 0.86 0.85 428
Let us observe that the "Pos" and "Neg" labels do not match the actual confusion matrix values.
The true and correct confusion matrix output should be as follows:
Confusion matrix on set provided: -------------------------------------------------------------------------------- Predicted: Pos Neg -------------------------------------------------------------------------------- Actual: Pos 64 (tp) 40 (fn) Neg 20 (fp) 304 (tn)
Description:
The confusion matrix printout has misaligned "Pos" and "Neg" labels under the "Predicted" and "Actual" sections. This leads to incorrect associations between the labels and the corresponding confusion matrix values.
The
return_metricsgenerates the following confusion matrix using the AIDS Clinical Trials Notebook as an example:Let us observe that the "Pos" and "Neg" labels do not match the actual confusion matrix values.
The true and correct confusion matrix output should be as follows: