Skip to content

Commit

Permalink
Add code comments for documentation
Browse files Browse the repository at this point in the history
  • Loading branch information
yyysjz1997 authored Nov 19, 2023
1 parent 312d7ad commit e50de07
Show file tree
Hide file tree
Showing 2 changed files with 164 additions and 43 deletions.
183 changes: 152 additions & 31 deletions deepod/metrics/_anomaly_detection.py
Original file line number Diff line number Diff line change
Expand Up @@ -6,15 +6,71 @@


def auc_roc(y_true, y_score):
"""
Calculates the area under the Receiver Operating Characteristic (ROC) curve.
Args:
y_true (np.array, required):
True binary labels. 0 indicates a normal timestamp, and 1 indicates an anomaly.
y_score (np.array, required):
Predicted anomaly scores. A higher score indicates a higher likelihood of being an anomaly.
Returns:
float:
The score of the area under the ROC curve.
"""

return metrics.roc_auc_score(y_true, y_score)


def auc_pr(y_true, y_score):
"""
Calculates the area under the Precision-Recall (PR) curve.
Args:
y_true (np.array, required):
True binary labels. 0 indicates a normal timestamp, and 1 indicates an anomaly.
y_score (np.array, required):
Predicted anomaly scores. A higher score indicates a higher likelihood of being an anomaly.
Returns:
float:
The score of the area under the PR curve.
"""

return metrics.average_precision_score(y_true, y_score)


def tabular_metrics(y_true, y_score):
"""calculate evaluation metrics"""
"""
Calculates evaluation metrics for tabular anomaly detection.
Args:
y_true (np.array, required):
Data label, 0 indicates normal timestamp, and 1 is anomaly.
y_score (np.array, required):
Predicted anomaly scores, higher score indicates higher likelihoods to be anomaly.
Returns:
tuple: A tuple containing:
- auc_roc (float):
The score of area under the ROC curve.
- auc_pr (float):
The score of area under the precision-recall curve.
- f1 (float):
The score of F1-score.
"""

# F1@k, using real percentage to calculate F1-score
ratio = 100.0 * len(np.where(y_true == 0)[0]) / len(y_true)
Expand All @@ -27,12 +83,62 @@ def tabular_metrics(y_true, y_score):


def ts_metrics(y_true, y_score):
"""calculate evaluation metrics for time series anomaly detection"""
"""
Calculates evaluation metrics for time series anomaly detection.
Args:
y_true (np.array, required):
Data label, 0 indicates normal timestamp, and 1 is anomaly.
y_score (np.array, required):
Predicted anomaly scores, higher score indicates higher likelihoods to be anomaly.
Returns:
tuple: A tuple containing:
- roc_auc_score (float):
The score of area under the ROC curve.
- average_precision_score (float):
The score of area under the precision-recall curve.
- best_f1 (float):
The best score of F1-score.
- best_p (float):
The best score of precision.
- best_r (float):
The best score of recall.
"""

best_f1, best_p, best_r = get_best_f1(y_true, y_score)
return auc_roc(y_true, y_score), auc_pr(y_true, y_score), best_f1, best_p, best_r


def get_best_f1(label, score):
"""
Return the best F1-score, precision and recall
Args:
label (np.array, required):
Data label, 0 indicates normal timestamp, and 1 is anomaly.
score (np.array, required):
Predicted anomaly scores, higher score indicates higher likelihoods to be anomaly.
Returns:
tuple: A tuple containing:
- best_f1 (float):
The best score of F1-score.
- best_p (float):
The best score of precision.
- best_r (float):
The best score of recall.
"""

precision, recall, _ = metrics.precision_recall_curve(y_true=label, probas_pred=score)
f1 = 2 * precision * recall / (precision + recall + 1e-5)
best_f1 = f1[np.argmax(f1)]
Expand All @@ -43,38 +149,53 @@ def get_best_f1(label, score):

def ts_metrics_enhanced(y_true, y_score, y_test):
"""
Compared with ts_metrics, this function can return more metrics
with one more input y_test (predictions of events)
revised by @Yiyuan Yang 2023/11/08
This function calculates additional evaluation metrics for time series anomaly detection. It returns a variety of metrics, including those sourced from the code in [A Huet et al. KDD22] and [J Paparrizos et al. VLDB22]. The function requires three inputs: y_true (data label), y_score (predicted anomaly scores), and y_test (predictions of events).
Args:
y_true:
y_score:
y_test
y_true (np.array):
Data label, where 0 indicates a normal timestamp and 1 indicates an anomaly.
y_score (np.array):
Predicted anomaly scores, where a higher score indicates a higher likelihood of being an anomaly.
y_test (np.array):
Predictions of events, where 0 indicates a normal timestamp and 1 indicates an anomaly.
Returns:
auroc:
aupr:
best_f1:
best_p:
best_r:
affiliation_precision:
affiliation_recall:
vus_r_auroc:
vus_r_aupr:
vus_roc:
vus_pr:
Example:
from deepod.models.time_series import DCdetector
clf = DCdetector()
clf.fit(X_train)
pred, scores = clf.decision_function(X_test)
from deepod.metrics import point_adjustment
from deepod.metrics import ts_metrics_enhanced
adj_eval_metrics = ts_metrics_enhanced(labels, point_adjustment(labels, scores), pred)
print('adj_eval_metrics',adj_eval_metrics)
tuple: A tuple containing:
- auroc (float):
The score of the area under the ROC curve.
- aupr (float):
The score of the area under the precision-recall curve.
- best_f1 (float):
The best score of F1-score.
- best_p (float):
The best score of precision.
- best_r (float):
The best score of recall.
- affiliation_precision (float):
The score of affiliation precision.
- affiliation_recall (float):
The score of affiliation recall.
- vus_r_auroc (float):
The score of range VUS ROC.
- vus_r_aupr (float):
The score of range VUS PR.
- vus_roc (float):
The score of VUS ROC.
- vus_pr (float):
The score of VUS PR.
"""

best_f1, best_p, best_r = get_best_f1(y_true, y_score)
Expand Down
24 changes: 12 additions & 12 deletions deepod/metrics/_tsad_adjustment.py
Original file line number Diff line number Diff line change
Expand Up @@ -5,20 +5,20 @@ def point_adjustment(y_true, y_score):
"""
adjust the score for segment detection. i.e., for each ground-truth anomaly segment,
use the maximum score as the score of all points in that segment. This corresponds to point-adjust f1-score.
** This function is copied/modified from the source code in [Zhihan Li et al. KDD21]
*This function is copied/modified from the source code in [Zhihan Li et al. KDD21]*
Parameters
----------
y_true: np.array, required
data label, 0 indicates normal timestamp, and 1 is anomaly
Args:
y_true (np.array, required):
Data label, 0 indicates normal timestamp, and 1 is anomaly.
y_score (np.array, required):
Predicted anomaly scores, higher score indicates higher likelihoods to be anomaly.
y_score: np.array, required
predicted anomaly scores, higher score indicates higher likelihoods to be anomaly
Returns
-------
score: np.array
adjusted anomaly scores
Returns:
np.array:
Adjusted anomaly scores.
"""
score = y_score.copy()
Expand Down

0 comments on commit e50de07

Please sign in to comment.