API Reference

Every public function and class, grouped by the stage of the pipeline it belongs to.

Detectors share one interface, described in BaseDetector: fit / decision_function / predict / is_anomaly, with decision_scores_, threshold_ and labels_ set after fitting. Classical and deep detectors are therefore interchangeable.

Data

• Ingestion
  • load_smap_labels()
  • anomaly_point_mask()
  • load_smap_channel()
  • load_smap()
  • load_from_csv()
  • is_long_form()
  • coerce_long()
  • pick_time_column()
  • load_from_ccsds()
  • TelemetryDataset
• Preprocessing Module
  • clean()
  • dedupe()
  • integrity_check()
  • resample()
  • interpolate_gaps()
  • normalize_fit()
  • pipeline()
• Feature Extraction Module
  • pivot_wide()
  • windowify()
  • features_stat()
  • make_feature_table()

Detection

• BaseDetector
  • Design
  • Input Convention
  • Post-fit Attributes
  • BaseDetector
• Unsupervised Models
  • PCAAnomaly
  • KMeansAnomaly
  • IsolationForestModel
• Deep Models
  • GDN
  • KANGDN
  • Distillation
  • C Code Generation
• Ensemble Model
  • AnomalyEnsemble

Scoring

• Thresholding Module
  • anomalous_sequences()
  • dynamic_threshold()
  • threshold_for_budget()
  • filter_sequences()
  • prune_sequences()
  • detect_anomalies()
• Evaluation Module
  • point_adjust()
  • prf()
  • point_adjusted_f1()
  • windows_to_points()
  • windows_to_point_scores()
  • best_point_adjusted_f1()
  • pr_auc()
  • false_alarm_rate_at_recall()
  • evaluate_sequences()
  • f_beta()
  • sequence_prf()