Anomaly detection is the process of identifying the data deviation from the expected results in a time-series data. This deviation can have a huge impact on forecasting models if not identified before the model creation. Snowflake Cortex AL/ML suite helps you train the models to spot and correct these outliers in order to help improve the quality of your results. Detecting outliers also helps in identifying the source of the deviations in processes.

Anomaly detection works with both single and multi-series data. Multi-series data represents multiple independent threads of events. For example, if you have sales data for multiple stores, each store’s sales can be checked separately by a single model based on the store identifier. These outliers can be detected in time-series data using the Snowflake built-in class SNOWFLAKE.ML.ANOMALY_DETECTION.