Introduction
Amazon Lookout for Metrics is a new facility provided by Amazon that detects outliers in time-series data, determines their causes, and allows us to act promptly. Without any machine learning knowledge, we can use Amazon Lookout for Metrics to create highly accurate machine learning models called Detectors to discover outliers in the data. Outliers in live or real-time data are detected using Amazon Lookout for Metrics. Amazon Lookout for Metrics will utilize historical data to build a model to detect outliers in real-time data if historical data is available. If no previous information is available, Amazon Lookout for Metrics will train a model on the fly. We provide Amazon Lookout for Metrics with the location and scope of our data via the Amazon Lookout for Metrics dashboard or SDK. This contains the Measures, which are the variables we want to look at, such as revenue, and the Dimensions, which are categorical variables that correspond to a measure.
Amazon Lookout automatically selects the optimum machine learning algorithm for our outlier detection use case for Metrics, which starts training a detector. Amazon Lookout uses these custom-trained detectors for Metrics to monitor our chosen metrics for outliers, allowing us to discover and rectify issues that may impact our business swiftly. Amazon Lookout for Metrics can also work with Amazon SNS to send us notifications when the service finds significant deviations.
Customers can get help from Amazon Lookout for Metrics to nineteen popular data sources. These include Amazon Storage Service, Amazon CloudWatch, Amazon Relational Database Service (RDS), Amazon Redshift, and SaaS applications. SaaS applications include Salesforce, Marketo, and Zendesk, to continuously monitor business-critical metrics, e.g., average purchase frequency, gross margin, return on advertising spend, etc.
Amazon Lookout for Metrics inspects and prepares the data automatically, then chooses the appropriate machine learning algorithm for the job, starts finding abnormalities, combines related anomalies, and highlights likely root causes. If a customer's website traffic unexpectedly drops, Amazon Lookout for Metrics can rapidly identify whether the reason is an unintended deactivation of a marketing campaign. Customers can also prioritize which issue to address by ranking the anomalies by projected severity. Customers can build customized alerts or actions like filing a trouble ticket or removing an erroneously priced product from a retail website by connecting Amazon Lookout for Metrics. Lookout for Metrics is available through the AWS console and through AWS Partner Network supporting partners who may assist clients in developing custom solutions using the service. The solution also works with AWS CloudFormation and complies with the General Data Protection Regulation of the European Union.
How It Works
Unlike more standard rule-based models, Amazon Lookout for Metrics examines KPIs with a context-rich eye to identify any unexpected trend-averse deltas and determine the core cause of an anomaly. When a sudden change in the rate of shopping cart abandonment, payment transaction failures, or sales revenue occurs, Amazon Lookout for Metrics detects the change, generates an alert for the organization, and diagnoses potential root causes to help decision-makers address problems faster. Suppose Amazon Lookout for Metrics is a product version of a machine learning-based KPI monitoring and management solution Amazon built to solve its operational challenges. Amazon can innovate and monetize itself. This is an example after AWS products have matured into commercial services.
Also see, Amazon Hirepro
137+ registered 
