What is the 'Cold Start' problem in anomaly detection?

Updated May 5, 2026

Short answer

When a new system or user has no historical data for a baseline[cite: 1].

Deep explanation

Without history, the model cannot distinguish between a new user's valid behavior and an anomaly[cite: 1].

Real-world example

A new bank account where any large transaction might look suspicious[cite: 1].

Common mistakes

  • Flagging all new users as anomalies[cite: 1].

Follow-up questions

  • How to solve it?

More Anomaly Detection interview questions

View all →