d-ailin / GDN

Implementation code for the paper "Graph Neural Network-Based Anomaly Detection in Multivariate Time Series" (AAAI 2021)
MIT License
481 stars 141 forks source link

some question about the auc score #95

Closed 0nyoun9 closed 6 months ago

0nyoun9 commented 6 months ago

Hi, thanks for your excellent work! I saw in your code that you calculate the AUC score using gt_labels and total_topk_err_scores. auc_score = roc_auc_score(gt_labels, total_topk_err_scores) Can I use gt_labels and pred_labels to calculate the AUC score

d-ailin commented 6 months ago

When you already have pred_labels (0 or 1), you should evaluate other metrics, like precision instead of AUC. AUC can be seen as an average performance metric under different thresholds (these thresholds will decide 0 or 1 labels).

0nyoun9 commented 6 months ago

When you already have pred_labels (0 or 1), you should evaluate other metrics, like precision instead of AUC. AUC can be seen as an average performance metric under different thresholds (these thresholds will decide 0 or 1 labels).

i get it,thank you!