Design, deliver, and run enterprise blockchain workloads quickly and easily.
How can machine learning help your organization?
What’s the next big technological advance for businesses? The cloud? Big data? Machine learning? Or maybe a combination of all three that actually generates business value?
That last one is surprisingly tricky. Most Global 2,000 enterprises lack the ability to leverage business data strategically via techniques like consolidation, analytical abstraction, and placing machine-learning systems on top of their data.
Machine learning involves building applications that eliminate the need for human intervention by sifting through your data and making smart decisions for your business. Examples include systems that automatically replenish inventory based on weather patterns and historical trends, or that optimize truck routes using Google API data on traffic.
The trick is to move machine learning and other analytic processes from the domain of R&D into the business. That means moving the understanding of data—including predictive analytics—from executive dashboards to core business processes. Doing so allows machine learning capabilities, such as inventory depletion predictions or least-cost routing, to be accessed directly from the software that companies use to run their business.
Leveraging the cloud
The problem with using a fairly new technology such as machine learning is that most people don’t understand its value until they see it used to solve a real-life problem. Keep in mind that artificial intelligence (AI), which is the basis of machine learning, has been around for decades, and machine learning is really just a more practical application of AI concepts.
So what has changed? For one thing, the rise of cloud computing. Machine learning systems are very expensive to build and operate—so much so that the application of machine learning technology must be carefully selected. Historically, machine learning was rarely used in business because few enterprises found good use cases for it.
But businesses can now access cloud-based machine learning services from the likes of AWS, Google, and Microsoft for pennies a minute. They also benefit from cheap cloud storage for data and associated machine learning systems.
So, now that it’s affordable, what types of use cases are out there? The combination of machine learning, big data, and the cloud promises to have a significant impact in industries where there is room for more efficiency, such as healthcare, transportation and retail.
Machine learning for retailers
The application of machine learning in the retail space offers many strategic advantages. Most retail companies operate on very thin margins, so any ability to boost efficiency will likely go right to their bottom lines.
Retailers can now analyze existing data to understand sales trends over a year. They can also leverage outside information such as weather data or key economic indicators. Machine learning systems that are bolted on to these data sources can make key decisions based upon the latest information available.
Of course, predictive analytics does not require machine learning. But adding a machine learning system to the analytical engine allows the system to learn over time. Thus, sales predictions that drive buying patterns as well as inventory are made from existing internal and outside data. The machine learning system adds the ability to understand what the data really means in the context of many data patterns.
Unlike humans, machine learning systems can consider thousands of patterns in the data and connect those patterns to likely outcomes. For instance, the system can figure out that a given brand is more likely to sell during presidential election years or when the weather is trending warmer. The result for retailers is higher sales and lower costs.
Machine learning for healthcare
In healthcare, we find similar decision-making patterns based on patient data. Wearable devices, for example, can spin off thousands of data points a minute. However, this data is meaningless unless it’s constantly analyzed. It’s not a good use of a clinician’s time to spend hours staring at a computer screen, looking for data patterns that might predict a negative health event.
Machine learning systems can pull wearable data streams and compare current and past levels for blood pressure, heart rate, physical activity, blood oxygen, etc. This data is analyzed against historical patterns for thousands of other patients and include outcomes of those data patterns, such as a heart attack or stroke. Using this information, the machine learning system can understand when your data indicates a problem. The system considers your past outcome data, as well as current medications, family medical history and genetic predispositions. If there is an issue, both you and your doctor are alerted.
Physicians can use this kind of information to guide treatment decisions. It can also be embedded in business processes, such as clinical systems that schedule medical procedures based on urgency. The result is lower patient mortality.
Machine learning for logistics
Machine learning can also be used to solve logistical problems. For example, a liquor distributor might use Google’s traffic API and data from Weather.com to optimize delivery routes.
In this case, the machine learning system takes only three data sets into account: traffic, weather, and route data. The idea is to find the ideal path for the delivery vehicles based upon experience that’s built into the machine learning system over years. Data also considered would be load data and deliveries that had to be made that day, including location.
The “best route” is then transmitted to the drivers in real time as they move from delivery to delivery. Updates can be made automatically to adjust for weather and traffic changes. The result could be significant savings in fuel and maintenance costs.
Systems that learn
Here's the real question: How can machine learning help your organization?
Much depends on your use of data. Machine learning systems are only as good as the data that drives them. So step one is to break down data silos. Step two is data consolidation. The use of public clouds can dramatically lower the cost and the risk of doing so, but you must understand the cost and risk for your specific data and usage patterns.
Data-driven machine learning systems take a lot of skill to set up, and that is where the majority of costs are found. But if you find the right data set to boost you company's efficiency, that investment can yield big rewards down the line.
Machine learning: Lessons for leaders
- Machine learning uses your data to make smart decisions for your business.
- Combine machine learning, big data, and the cloud to create efficiencies in industries such as healthcare, transportation and retail.
- The use of public clouds can dramatically lower the cost of data consolidation.
This article/content was written by the individual writer identified and does not necessarily reflect the view of Hewlett Packard Enterprise Company.