paulpas Anomaly Detection
Comprehensive guide to anomaly detection in machine learning and data science workflows.
When to Use This Skill
- Solving real-world unsupervised learning problems
- Building machine learning pipelines with anomaly detection
- Implementing best practices for anomaly detection
- Optimizing model performance using anomaly detection techniques
- Learning industry-standard approaches to anomaly detection
When NOT to Use This Skill
- When using pre-built libraries without understanding underlying concepts
- For toy problems that don't require anomaly detection rigor
- When domain expertise in specific problem requires different approach
- If your problem doesn't require the complexity this skill provides
Purpose and Key Concepts
Anomaly Detection is a critical component of the machine learning workflow. This skill covers:
- Theoretical foundations — Mathematical principles and statistical concepts
- Practical implementation — Working code examples and patterns
- Common pitfalls — Mistakes to avoid and how to recover from them
- Best practices — Industry-standard approaches and optimization techniques
Core Workflow
- Understand the problem — Clearly define what you're solving for
- Select approach — Choose the right technique for your data and constraints
- Implement solution — Write clean, tested code following best practices
- Validate results — Verify your implementation with tests and validation
- Optimize performance — Improve efficiency and accuracy incrementally
Implementation Patterns
Pattern 1: Basic Anomaly Detection
import pandas as pd
import numpy as np
from sklearn.ensemble import IsolationForest
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
def basic_anomaly_detection(data: pd.DataFrame, contamination: float = 0.1) -> dict:
"""Perform basic anomaly detection using Isolation Forest."""
if data.empty:
raise ValueError("Input DataFrame cannot be empty")
X = data.values
X_train, X_test = train_test_split(X, test_size=0.2, random_state=42)
model = IsolationForest(contamination=contamination, random_state=42, n_estimators=100)
model.fit(X_train)
train_scores = model.decision_function(X_train)
test_scores = model.decision_function(X_test)
test_labels = model.predict(X_test)
return {
"train_scores": train_scores
"test_scores": test_scores
"test_labels": test_labels
"model": model
}
Pattern 2: Production-Ready Anomaly Detection
import logging
import pandas as pd
import numpy as np
from typing import Any, Dict, List
from sklearn.ensemble import IsolationForest
from sklearn.neighbors import LocalOutlierFactor
from sklearn.preprocessing import StandardScaler
logger = logging.getLogger(__name__)
class AnomalyDetection:
"""Production implementation of Anomaly Detection supporting multiple algorithms.
Adheres to SOLID principles for maintainability and testability."""
def __init__(self, algorithm: str = "isolation_forest", contamination: float = 0.1):
if algorithm not in ["isolation_forest", "lof"]:
raise ValueError("Algorithm must be 'isolation_forest' or 'lof'")
self.algorithm = algorithm
self.contamination = contamination
self.scaler = StandardScaler()
self.model = None
def execute(self, data: pd.DataFrame) -> Dict[str, Any]:
"""Execute Anomaly Detection on data"""
if data.empty:
raise ValueError("Input data cannot be empty")
X = data.values
X_scaled = self.scaler.fit_transform(X)
if self.algorithm == "isolation_forest":
self.model = IsolationForest(
contamination=self.contamination
random_state=42
n_estimators=200
)
else:
self.model = LocalOutlierFactor(
n_neighbors=20
contamination=self.contamination
novelty=True
)
self.model.fit(X_scaled)
scores = self.model.decision_function(X_scaled)
labels = self.model.predict(X_scaled)
logger.info(f"Anomaly detection completed. Found {np.sum(labels == -1)} anomalies.")
return {
"scores": scores
"labels": labels
"anomaly_count": int(np.sum(labels == -1))
"anomaly_ratio": float(np.mean(labels == -1))
"model_type": self.algorithm
}
Best Practices
- ✅ Always validate your implementation on test data
- ✅ Document your assumptions and methodology
- ✅ Use version control for reproducibility
- ✅ Monitor performance metrics in production
- ✅ Periodically review and update your approach
- ✅ Test with edge cases and outliers
- ✅ Log all significant operations for debugging
Common Pitfalls
| Pitfall | Problem | Solution | |
Constraints
MUST DO
- Validate all data preprocessing steps are fit-only on training data, never on validation or test sets
- Implement reproducible pipelines with fixed random seeds and deterministic operations where possible
- Report model performance with confidence intervals via bootstrapping or cross-validation across multiple runs
- Log all experiments with parameters, metrics, and artifacts using MLflow or equivalent tracking system
MUST NOT DO
- Do not evaluate a model on the same data used for training — always hold out a proper test set
- Avoid overfitting to the validation set by limiting hyperparameter search iterations
- Never use features that can only be computed at inference time (look-ahead bias)
- Do not report single-run accuracy without statistical significance testing or error bars
Live References
Authoritative documentation links for this skill's domain. The model follows markdown links at load time to resolve external references and inline content.