The algorithm correlates cross-siloed revenue-impacting anomalies across the enterprise, speeding time to detection and remediation by 30%
Anodot, the autonomous business monitoring company, announced that it had been granted the US patent US10891558B2 for its Heuristic Inference of Topological Representation of Metric Relationships. The patent covers Anodot’s first-of-its-kind machine learning (ML) based correlation analysis that allows enterprises to automatically discover non-obvious cross-siloed anomalies that can impact revenues.
Correlation analysis identifies relationships between key performance indicators, which business teams can quickly use to determine the root cause revenue and service impacting events. Without Anodot, companies would have to use time-consuming manual techniques to determine the revenue disruption’s underlying cause. By combining anomaly detection with automated cross-siloed correlation analysis, Anodot helps enterprises detect and fix revenue-impacting incidents 80% faster than any other method.
“Historically, events that have produced lost revenue or service drops were difficult to detect because the correlating events could be in different places within the enterprise requiring different domain expertise. For example, the team in charge of revenue could notice a drop in sales for a particular selection of items but wouldn’t know that the cause of the problem was the slow page load times for items with overly large image sizes,” said Anodot Chief Data Scientist Dr. Ira Cohen. “Using Anodot’s correlation analysis, the revenue team, which mostly likely does not have in-depth knowledge of web page design, would immediately be alerted to the problem, determine its cause, be pointed to a solution, and minimize revenue loss.”
The power of the Anodot platform is its ability to accurately detect and correlate anomalies at scale without the need for any human input. Anodot invented an algorithm called abnormal correlation that groups metrics across silos if they behave abnormally at similar times. The theory is that if two metrics are affected by a common cause, companies should see that pattern repeat itself over time. Anodot uses locality-sensitive hashing (LSH) to scale its correlation techniques to sift through billions of metrics. The combination of abnormal correlation and LSH provide Anodot customers with the fastest time to remediation for events that can impact service and revenue.
“Anodot goes beyond mere alerts,” said Viacheslav Tsyganov, Chief Information Officer, Vice President, Deputy Chairman of the Management Board at Tinkoff. “Anodot’s correlation analysis not only accurately alerts us when there is a disruption in service, it also shows us why the problem is happening. Because the root cause analysis helps us understand the relationships between anomalies across teams and departments, we fix problems faster and suffer fewer revenue losses.”
For more such updates and perspectives around Digital Innovation, IoT, Data Infrastructure, AI & Cybsercurity, go to AI-Techpark.com.