Glossary · Business Analytics
Anomaly Detection
also: outlier detection · isolation forest
Definition
Anomaly detection identifies observations that deviate meaningfully from expected behavior, accounting for trend, seasonality, and variance. In revenue data it separates true incidents (payment outages, pricing bugs) from normal fluctuation. Isolation forests and Prophet-based decomposition are the practical workhorses.
Simple z-score thresholds fail on real business data because variance is non-constant and seasonality is multi-layered (daily, weekly, holiday). Prophet-based decomposition fits trend + seasonality and flags residuals beyond an uncertainty interval. Isolation forests score anomalousness based on how easy a point is to isolate in random partitioning — a model-free, density-based approach that handles mixed feature types.
Essays on this concept
- Business Analytics
Anomaly Detection in Revenue Data: Isolation Forests vs. Prophet-Based Decomposition
A 4% revenue drop on a Tuesday could be a payment processor outage, a pricing bug, or just normal variance. The difference between these explanations is millions of dollars — and your monitoring system can't tell them apart.
- Business Analytics
From Dashboards to Decision Systems: Embedding Prescriptive Analytics Into Operational Workflows
Your company has 47 dashboards. How many of them changed a decision last week? Dashboards describe what happened. Decision systems prescribe what to do next — and the gap between these two is where most analytics ROI evaporates.
- E-commerce ML
Demand Forecasting with Conformal Prediction: Reliable Uncertainty Intervals for Inventory Optimization
Your demand forecast says you'll sell 1,000 units next month. How confident is that prediction? Traditional models give you a number without honest uncertainty bounds. Conformal prediction gives you intervals with mathematical coverage guarantees — no distributional assumptions required.
- Marketing Strategy
Market Sensing Systems: Building an Automated Competitive Intelligence Pipeline with LLMs and Structured Data
Your competitor raised prices three weeks ago. Changed their positioning last month. Started hiring ML engineers in Q3. You found out in a strategy meeting yesterday. Automated market sensing closes this gap from weeks to hours.
- E-commerce ML
Real-Time Fraud Detection at Checkout: A Streaming ML Pipeline Architecture with Sub-100ms Latency
You have 100 milliseconds to decide whether a transaction is fraudulent. In that window, you need to compute 200+ features from streaming data, run inference on a model trained on 1:1000 class imbalance, and return a score that balances revenue loss against customer friction.
Related concepts