What are extreme data events?

Extreme data events, also known as outliers, are rare occurrences in data that can heavily skew high-level metrics or negatively impact business, such as a bug allowing negative shopping carts or a competitor scraping a site.

Why don't traditional outlier detection approaches work well for real-world data?

Traditional approaches often fail because real-world data has many columns in various formats, changes over time, contains irrelevant outliers, and can be challenging to interpret, especially with large datasets.

How does Anomalo's Entity Outlier check identify outliers?

Anomalo's approach narrows the scope by requiring an 'entity' (e.g., IP address) and a 'metric' (e.g., site visits). It then applies time series anomaly detection and root cause analysis techniques to provide reliable, clear, and relevant outlier detection.

Can Anomalo explain why a specific data point is an outlier?

Yes, Anomalo automatically profiles complaints about an address and compares them to other incidents to identify segments that characterize them, providing rich and actionable context for users.

Blog

Detecting Extreme Data Events: Anomalo's Innovative Outlier Detection

November 16, 2021

Home
Blog
An Innovative Approach to Detecting Extreme Data Events

Anomalo's innovative approach to detecting extreme data events focuses on identifying rare, impactful occurrences that can skew high-level metrics or negatively impact business operations, such as fraudulent transactions or competitor scraping. This automated solution addresses the limitations of traditional outlier detection methods by narrowing the scope to specific entities and metrics, then applying time series anomaly detection and root cause analysis for reliable, clear, and relevant outlier identification in vast datasets.