Machine Learning: The Foundation for New-Age Predictive Analytics
- By Sameer Nori
- January 11, 2017
There used to be a time when relying on reports based on historical data was the norm and considered acceptable for making business decisions. Statistics and forecasting techniques helped augment those decisions but never quite took off for a variety of reasons, including the need to run forecasts based on sample sizes and the inability to incorporate new data sources. The initial promise of big data was to fulfill these unmet requirements -- and finally, the industry and ecosystem have made significant progress with machine learning-based predictive analytics which is starting to hit its stride.
By using predictive analytics, organizations can find patterns in historical and transactional data to identify risks and opportunities yet unfolding. What's even better is that current real-time streams of data can also be incorporated into this analysis. That's really where the power of machine-learning predictive analytics can be felt. Let's look at three scenarios.
Increasing Top-Line Revenue
Marketing and sales are probably the two functional areas that are undergoing the greatest transformation. This shift was first observed within high-tech companies but has made its way into other industries as well, such as pharmaceuticals, CPG, and retail. Traditional methods for reaching customers in both the B2B and B2C space are no longer reliable. The explosion of channels and the amount of research that customers do online about products and services is causing marketing and sales functions to rethink their customer acquisition strategy.
Enter predictive analytics that enables scoring of individual customers across all stages of their buying journey. This metric is accomplished by storing and analyzing clickstreams, sales data, responses to campaigns, and sentiment expression on social media. Changes to online Web properties as well as adjustments to promotions and campaigns can happen rapidly by utilizing machine learning-based predictive analytics. The result is a focus on customers that matter the most and higher ROI for campaigns overall. It's the early days in the development of artificial intelligence, but it's fair to say that AI is going to have its impact on this space as well.
You're probably tired of hearing the phrase "do more with less." The reality, though, is that every organization has to figure out how to get maximum mileage from its existing assets and infrastructure and find ways to realize operational improvement. Let's think about this truth in the context of the oil and gas industry and consider how predictive analytics might apply. Being able to predict when existing oil rigs or other equipment need overhaul and maintenance can shave off significant costs, regardless of where they are in the value chain. This forecast can be done by tapping into the enormous amount of data that comes from sensors embedded in every piece of equipment and then marrying that information with historical data about the maintenance life cycle as well as other sources. The resulting cost savings can be used to find new areas of growth rather than spent on existing assets and infrastructure.
Improving Patient Outcomes and Lives
The overall healthcare industry is one that probably has the greatest ability to tap into the promise of machine-learning-based predictive analytics. There are various players in the overall value chain that stand to benefit -- in this instance, we're going to look at it through the lens of a hospital. Storing and analyzing vast amounts of patient data from wearable devices, past medical history, previous prescriptions, and home health devices, among other sources, presents a treasure chest full of opportunities. Hospitals can benefit from predictive analytics served up from this data, enabling them to detect diseases earlier and quicker, reduce re-admittance rates, and improve resource allocation in nursing units. This example is only the tip of the iceberg when it comes to the vast applications of machine learning-based predictive analytics.
Certainly, machine-learning predictive analytics might seem daunting if you think about the technical details involved to get this process going. The good news is that tremendous progress has been made on the tools and ecosystem surrounding it, significantly improving its usability. This user-friendly development is making it far easier for business people who are not IT or data scientists to get in on the action themselves and try self-service with machine learning-enabled predictive analytics. There's certainly more to be done to enable this class of analytics to be available to everyone, much like a utility and on-demand service.
It is, however, a far cry from the analytics of old when executives and other business leaders were dependent on the IT department to get trending reports and other vital information. Obtaining the final output could take weeks or even months, and by the time the information was available, the initial ask was no longer valid. With today's solutions, there is no need to burden IT when undertaking a predictive analytics task and this evolution promises to get better. As Yogi Berra said, "The future ain't what it used to be."
Sameer Nori is senior product marketing manager at MapR. In his career, he has worked at companies including SAP Business Objects, MicroStrategy, and Jaspersoft, where he honed his expertise in business intelligence, analytics, and big data markets. You can reach the author at firstname.lastname@example.org.