Time series anomaly detection is a tool that detects unusual behavior, whether it's hurtful or advantageous for the business. In either case, quick outlier detection and outlier analysis can enable you to adjust your course quickly, before you lose customers, revenue, or an opportunity. The first step is knowing what types of outliers you’re up against.

Chief Data Scientist Ira Cohen, co-founder of Autonomous Business Monitoring platform Anodot, covers the three main categories of outliers and how you'll see them arise in a business context:

• Global Outliers (aka Point Anomalies)
• Contextual Outliers (aka Conditional Anomalies)
• Collective Outliers

For more on machine learning-based anomaly detection, visit www.anodot.com/blog and www.anodot.com/resources.