By Cassandra Balentine
The value of business data is well known today. Organizations understand the importance of big data and incite processes for capturing and managing data. Further, predictive analytics tools are implemented to drive artificial intelligence (AI) and machine learning with AI, expand the benefits of the Internet of Things (IoT), and improve customer experience.
Any business that has a use for data can benefit from analytics, and predictive technologies are no exception.
Predictive analytics is gaining traction across many industry verticals. Shawn Rogers, senior director, analytic strategy, TIBCO, suggests that it is popular amongst most use cases that leverage data for insights and actions, including financial services, healthcare, and manufacturing. “Industries that derive value from understanding historical views of their business or services can generally find greater value in a predictive view of their business,” he explains.
Himanshu Pande, senior director, product management, CA Technologies, says as businesses go digital, there is a realization that analytics are key for sustaining growth in a highly competitive world. “Often, knowing more than the competitor leads to success. This has caused a wide range of industries to become interested in predictive analytics, from banking to telecommunications and media to retail.”
Pande says the promise of gaining real-time visibility and insights into large, diverse sets of customer and systems data by applying the latest data science and machine learning techniques appeals to customer-centric companies. “Advances in technology, the open-source movement, and the infrastructure availability come together to make it economically viable for companies to embed advanced analytics into their applications,” he offers.
Lance Olson, partner director of program management, Cloud AI Platform, Microsoft, adds natural resources, retail, smart cities, and transportation to the list of early adopters. He points out that there are still a lot of businesses on the sidelines when it comes to using predictive analytics. “A recent Keystone Strategy report showed only ten percent of businesses will commercialize data by 2020. Our goal is to help all customers, wherever they are in their data and analytics journey and to build systems of intelligence and leverage data for business outcomes,” he says.
John Crupi, VP IoT Analytics, Greenwave Systems, says industrial power, industrial IoT, and consumer IoT verticals utilize predictive analytics. He says each of these verticals have time-sensitive information that requires timely actions. “Industrial IoT has always adopted the connected device, but now the move is into edge intelligence and predictive maintenance and analytics. “It’s too reactive and expensive to continue the traditional break/fix model. Instead, all organizations must use real-time patterns and analytics from devices combined with machine learning and AI techniques to understand machine behavior and predict future outcomes. This will dramatically reduce maintenance costs and optimize overall operations.”
Additionally, Olsen suggests that a business prepared to capture and analyze information is simply better at adapting and can move faster. “A recent Keynote Strategy study suggests companies that transform to a data-driven mindset are able to harness real-time data to deliver highly personalized customer experiences, optimize production runs based on forecasted demand, predict and minimize downtime, anticipate order fluctuations, empower employees with business intelligence and data visualization tools, and keep pace with the change.”
Predictive Analytic Use
Companies utilize predictive analytics to better optimize business processes, adding a deeper level of insight and predictive actions that add value. “Advanced analytics is becoming pervasive throughout many companies who are looking for greater competitive advantage,” says Rogers.
Customer-centric companies use predictive analytics to explore their customer interactions to deliver the best possible user experience and increase customer satisfaction. “One common use case includes improving online stores and services to give customers quick and easy access to personalized content. “Companies can then offer the exact goods and services, advice, and customer support a customer is looking for. Predictive analytics is also simplifying and enriching mobile experiences by predicting the customer journey and search patterns, proactively identifying network issues, provisioning capacity, and applying fixes without a need for human involvement, just to name a few,” says Pande.
In healthcare, Rogers says using predictive analytics can affect patient care. “Combining patient history data, real-time medical device data, and information from the operating room allows analytic platforms to predict the safest and most effective treatment for patient care, helping doctors reduce infections and provide the best care for patients.”
Microsoft’s Olson provides several cases of predictive analytics in use today by notable organizations.
For example, Rolls-Royce applied the predictive analytic capabilities of the Microsoft Azure IoT Suite to access data that helped it reduce fuel consumption, minimize maintenance costs, and improve the customer experience. This was implemented in an effort to better serve its customers and maintain its more than 13,000 commercial aircraft engines around the world.
Schneider Electric created edge intelligence with Microsoft Azure Machine Learning and Azure IoT Edge to transform its IoT solution to be better equipped to help customers protect their assets and the environment, while boosting workplace safety.
Weka Solutions applied the Azure IoT Suite to refrigerators to help field clinicians overcome unreliable vaccine transportation and storage practices. As a result, a Vaccine Smart Fridge uses remote monitoring and other IoT technology to better monitor, maintain, and automate life-saving vaccine storage and distribution—including remote areas around the globe where power is unreliable.
As BI continues to grow, opportunities for advanced analytics expand.
“Predictive analytics is more popular than ever,” says Rogers. “Companies are becoming more sophisticated in their application of predictive analytics and there is greater demand for these type of insights. A new community of consumers has arrived in the workplace who want to benefit from predictive insights and actions. They are creating a perfect storm of disruption as they demand access to predictive technology to better run their projects and departments,” he explains.
Olson points to AI as one major trend. “We’re seeing a significant increase in the use of AI in the cloud, edge, and mobile devices. Machine learning models are often built in the cloud and then deployed into the location that is closest to the point of decision. This results in a better experience, especially in cases where the time to decision is important, like on connected control systems or interactive mobile applications,” he explains.
In addition to AI used for machine learning, pre-built AI is used to accelerate solution deployment. “We’ve seen tremendous excitement in the use of pre-built models, accessible through our Cognitive Services, which enable developers to make use of models to solve problems working with vision, speech, language, knowledge, and search,” says Olson.
Conversational AI is also growing in popularity. “With advancements in key technologies like speech and language understanding, thousands of new conversational applications, bots, are being created each week in Microsoft Azure. Bots make services more accessible and engaging to people through a conversational interface that is more natural in many cases where a keyboard or mouse are less ideal.”
Crupi says democratizing AI and machine learning tools for non-PhDs is a current trend, as is CPU-based deep learning at the industrial edge.
“In the end, utilizing predictive analytics is all about serving your customers better and having an edge over your competitor. Technology companies are looking for a solution to their business problems, rather than abstract applications of the latest and greatest technologies,” suggests Pande. “A comprehensive set of data streams from multiple sources is more likely to help discover such solutions. Hence, the interest in analytics engines capable of ingesting data from a multitude of diverse sources quickly and easily. It also applies advanced algorithms to understand the full picture instead of looking at data from one angle.”
What to Watch
The future of predictive analytics is promising as are technologies fueled by its advancements.
Crupi says AI will be the biggest driver of predictive analytics. “Deep learning will take predictive analytics to new heights and begin democratizing it so adoption analysts and engineers can apply AI to solving problems in days and weeks, not months and years.”
Rogers says IoT cases will grow in popularity. “Edge analytics will become commonplace as more companies look to incorporate device-driven data and predictive analytics. IoT data has significant gravity and velocity, making it necessary for companies to take analytics to the data instead of always moving data to the analytics,” he offers. “Deploying analytics where the data is produced helps to overcome latency of decisions and the complexity of constantly moving data from the edge to the core of analytic environments.”
Rogers adds that deploying, managing, and optimizing predictive analytics across more complex and heterogeneous environments will challenge companies and their strategy for predictive analytics growth. “With increased use comes a need to more efficiently manage analytic environments,” he shares.
Pande says the application economy brings a growing interest in making every application analytics driven. “As a result there is growing demand for a set of skills commonly known as data science, including machine learning and AI across the industries.”
Predictive scenarios based on machine learning will become more pervasive. “We’ll see machine-learning driven advancements across nearly every domain and industry in the world, including healthcare, retail experiences, life sciences, finance, public services, manufacturing, transportation, and entertainment,” offers Olson.
Olson explains that most systems around us, such as machinery, infrastructure, and buildings will include sensors that are used to monitor the world around us and use data and machine learning models to predict potential future outcomes and take actions.
He says as computer vision and audio processing continue to advance, we’ll see far more sophisticated use cases that deploy relatively low-cost hardware, like a camera, to see and understand the world around them in much more meaningful ways. This might include the ability to look at a street and see if someone is standing in the street versus a person riding on a motorcycle or bike, and then determine whether or not it is safe to change the color of a stop light.
However certain challenges must be overcome to get to this point. Olson says one of the biggest challenges is processing data in a world where data growth rates accelerate exponentially. “Cloud platforms will play a critical role in achieving scale on this point,” he says.
Another challenge is building a system that can effectively manage, track, and retain the machine learning models deployed throughout the system. “Because machine learning is based on data from the world around us, and that world can change, the half-life of machine learning is much shorter than traditional software systems. We’ll need tools and services that understand how to measure and monitor the accuracy of predictive analytics up to hundreds of millions and determine variations in performance of the models and retrain and redeploy them automatically.
Crupi sees the biggest challenge as making AI a first-class priority in organizations. “This will require data scientists with deep learning expertise and domain expertise and engineers all working together. This must be supported from the highest levels of the organization,” he states.
Pandre says the time and resources it takes to hire and train a data scientist to full productivity is considered a high barrier for adoption of advanced analytics, such as predictive analytics.
A variety of predictive analytics platforms are found on the market today. Here are product highlights from solutions offered by vendors quoted in this article.
CA Jarvis is an analytics engine that receives data from applications, derives data from the data, and serves the data and insights back to the applications. The predictive analytics generated from Jarvis help applications deliver better customer experiences.
Greenwave Systems features its AXON Predict platform, which is a real-time, edge analytics platform. It includes built-in analytics, patterns, and highly-interactive visual analytics that provide real-time insight and actions for commercial and industrial IoT devices. AXON Predict is used for real-time and historical analytics and pattern detection to identify and predict machine behaviors with the goal of optimizing performance and avoiding unscheduled maintenance.
Microsoft offers big data and machine learning solutions that allow customers to transform data into intelligent action. Notable updates to some of its most popular big data and analytics solutions in 2017 include Azure Machine Learning services, Machine Learning Server, the launch of Azure Databricks, and the launch of Azure IoT Edge.
TIBCO offers its Spotfire for predictive analytics. It features a natively embedded R engine that enables it to run popular open source statistical language directly inside any analytics. It provides point-and-click authoring for users of all levels and offers more advanced users the ability to run custom R models in the analysis.
The potential of business data is continually realized as analytic platforms continue to advance. From the ability to reflect on historical data, the capability to react to real-time data, and now to learn from previous behavior to predict likely outcomes and proactively strategize, BI and analytics brings companies into a new era.
Feb2018, Software Magazine