Time Series Anomaly Detection With “Real World” Data

Image from here.

Time series anomaly detection is a very hard problem, especially when working with “real world” data. To illustrate what I mean by “real world” data, let's say you are working with multiple clients and each client is running a different longitudinal study. The clients want you to help them detect anomalies in their data while the study is running, so they can potentially fix these anomalies and…