Question

A SOC analytics team wants to predict the expected volume of authentication requests for each hour of the coming week so capacity and staffing can be planned, and so that unusual spikes can later be flagged against the forecast. Historical hourly authentication counts spanning two years are available. Which AI/ML technique should the team recommend to produce these numeric hourly predictions?

Accepted Answer

Time-series regression (forecasting) trained on the historical hourly counts

Answer

K-means clustering to group similar authentication events

Answer

Binary classification trained on labeled malicious/benign logins

Answer

Association rule mining to find co-occurring login attributes