1  """ADEngine: Full anomaly detection lifecycle.
 2  
 3  Demonstrates PyOD's ADEngine for automatic detector selection,
 4  execution, analysis, explanation, and report generation.
 5  """
 6  from pyod.utils.ad_engine import ADEngine
 7  from pyod.utils.data import generate_data
 8  
 9  # Generate sample data
10  X_train, X_test, y_train, y_test = generate_data(
11      n_train=300, n_test=100, n_features=20, contamination=0.1)
12  
13  # Initialize the engine
14  engine = ADEngine()
15  
16  # === Full lifecycle ===
17  print("=" * 60)
18  print("FULL ANOMALY DETECTION LIFECYCLE")
19  print("=" * 60)
20  
21  # 1. Profile
22  profile = engine.profile_data(X_train)
23  print("\n1. Data profile:", profile['data_type'],
24        "(%d samples, %d features)" % (profile['n_samples'],
25                                        profile['n_features']))
26  
27  # 2. Plan
28  plan = engine.plan_detection(profile, priority='speed')
29  print("2. Plan:", plan['detector_name'], "-", plan['reason'])
30  
31  # 3. Execute
32  result = engine.run_detection(X_train, plan, X_test=X_test)
33  print("3. Detection: %d anomalies (%.1f%%) in %.3fs"
34        % (result['n_anomalies'], result['anomaly_ratio'] * 100,
35           result['runtime_seconds']))
36  
37  # 4. Analyze
38  analysis = engine.analyze_results(result, X=X_train)
39  print("4. Analysis:", analysis['summary'])
40  
41  # 5. Explain
42  explanations = engine.explain_findings(result, X=X_train, top_k=3)
43  print("5. Top anomalies:")
44  for exp in explanations:
45      print("   Sample %d: score=%.4f (%s)"
46            % (exp['index'], exp['score'], exp['label']))
47  
48  # 6. Suggest next step
49  suggestion = engine.suggest_next_step(result, analysis)
50  print("6. Suggestion:", suggestion['action'], "-", suggestion['reason'])
51  
52  # 7. Report
53  report = engine.generate_report(result, analysis)
54  print("\n7. Report preview (first 500 chars):")
55  print(report[:500])
56  
57  # === Knowledge queries ===
58  print("\n" + "=" * 60)
59  print("KNOWLEDGE QUERIES")
60  print("=" * 60)
61  
62  print("\nAvailable text detectors:")
63  for d in engine.list_detectors(data_type='text'):
64      print("  %s: %s" % (d['name'], d['full_name']))
65  
66  print("\nECOD explained:")
67  info = engine.explain_detector('ECOD')
68  print("  %s" % info['full_name'])
69  print("  Best for: %s" % info['best_for'])
70  
71  print("\nADBench top 5:")
72  bench = engine.get_benchmarks('ADBench')
73  print("  %s" % bench['ADBench']['rankings']['overall_top_5'])