Follow Datanami:
March 1, 2018

Five Ways to Apply Streaming Analytics Now

Srinath Perera

Streaming analytics is moving into the mainstream as more enterprises capitalize on the technology to gain greater efficiency, competitiveness, and profitability. And as a result, the streaming analytics market will grow to $13.7 billion by 2021, at a compound annual growth rate of 34.8%, according to Markets and Markets.

Behind the demand are a number of emerging business uses for streaming analytics that are proving to deliver real-world benefits. Moreover, these uses are increasingly accessible as in-house and commercial enterprise solutions, alike, incorporate open source technologies for streaming analytics, such as Apache Storm and Apache Spark, and complex event processing (CEP) engines.

Many companies are familiar with the use of streaming analytics for sales and marketing applications, but there are five solutions for improving other areas of business that enterprises should consider in 2018 if they haven’t applied them already.

Predictive Maintenance

Whether managing manufacturing equipment or fleets of vehicles, enterprises can use streaming analytics to track wear, avoid failures, and schedule maintenance to minimize the impact on operations.

Here, a streaming analytics or CEP engine captures data from sensors about rate of wear, distance traveled, or other factors that can help predict when a part will have compromised performance. Applications combine this data with machine-learning models that can trigger alerts when maintenance needs to occur. This data-driven maintenance leads to major cost savings by minimizing downtime due to for example, a taxi’s worn-out alternator or a factory’s cracked machine part, as well as by avoiding overly conservative pre-set maintenance schedules.

Fraud Detection

Banks and other financial institutions have long relied on fraud detection, but the growth of IoT-enabled devices and online commerce is expanding its use across a range of industries.

Typically, normal behavior is modeled via machine learning using web and IoT events.  Streaming analytics are then combined with the model to detect deviations from those behavioral patterns. For instance, online shoppers typically spend time browsing on a website before going to checkout and paying. Streaming analytics can detect if multiple purchases are made with little or no web browsing time, which may suggest the presence of a bot. Similarly, streaming analytics can determine anomalies in the use of a smart device, such as a baby monitor, to determine if it has appropriated into a botnet.

Human Resources

Streaming analytics can help HR departments address the challenges of a tight job market by predicting churn and facilitating proactive intervention.

(Monkey Business Images/Shutterstock)

By applying streaming analytics to data streams, such as email, time reporting apps, injury reports, and other resources, managers can gain deeper insights into behavioral patterns that may suggest an employee is burning out due to excessive hours or actively interviewing at other companies. The insights can help HR professionals and line-of-business managers proactively balance workloads, offer more competitive compensation, or provide training and development to retain valued team members.

Meanwhile, patterns across work groups can alert management to safety, training or management issues that need to be addressed.

Customer Service

Streaming analytics, whether applied to the web or smart devices, can be used to track how a product is being used, provide customers with feedback, detect problems, and intervene and solve issues. For example, auto companies increasingly use sensors on their vehicles to identify and alert drivers to a developing issue, helping to ensure safety and avoid costly repairs. Alternatively, web data captured by streaming analytics may show that a customer is having difficulty with a software or mobile phone feature, which could trigger a chat box asking if the user would like help. On the flip side, streaming analytics help smart device providers share personalized stats with consumers, which lets them track performance and work toward goals.

Product and Service Innovation

Streaming analytics can produce unprecedented insights that are enabling companies to develop and sell new types of digital products and services they traditionally would not have recognized as opportunities. On the one hand, streaming analytics embedded into products can enhance or create new experiences, such as those in Bose headphones that let wearers talk in a noisy environment. Meanwhile, some companies are getting into the data business. They are realizing  that information collected with a streaming analytics or CEP engine—such as anonymized and aggregated credit card information, supermarket data, taxi ride trends, and highway videos—can be valuable to industries and policy makers. Similarly, Walmart shares such insights with partners and suppliers to help them improve their operations.

Whether companies are in the business of delivering products or services, streaming analytics solutions are offering opportunities to compete in today’s digital economy by driving innovation, building new revenue streams, growing customer loyalty, maximizing efficiency, and creating stronger teams. And by incorporating open source streaming analytics and CEP technology, many of these solutions are moving out of the bleeding edge and into helping businesses on main street.

About the Author: Dr. Srinath Perera is vice president of research in the CTO office at WSO2, Inc. He is responsible for documentation that offers insights into WSO2’s markets, and views on current and future technologies. Srinath is a scientist, software architect, and a programmer who works on distributed systems. He is a member of the Apache Software foundation and a key architect behind several widely used projects such as Apache Axis2 and WSO2 CEP. He has authored two books about MapReduce and is a frequent author of technical articles. Srinath received his Ph. from Indiana University, serves as a research scientist at the Lanka Software Foundation, and also teaches as a visiting faculty member at the Department of Computer Science and Engineering, University of Moratuwa.

Related Items:

Four Ways to Visualize Big Streaming Data Now

Merging Batch and Stream Processing in a Post Lambda World