A specialized piece of artificial intelligence, machine learning, allows computers to learn on their own. In “Why Machine Learning is Crucial to Effective Utilization of Big Data,” Cutter Consortium Senior Consultant Greg Smith and his co-authors explain that machine learning is a specialization within artificial intelligence (AI) that allows computers to learn on their own. Computers are provided a set of rules, rather than explicit programming instructions on what to do, and can then self-train and develop a solution by themselves. With ML techniques, computers iteratively learn from data and also find potential hidden insights in data.
Why is it that machine learning can offer a huge advantage over human intellect? According to Smith,
ML does not rely on human intelligence and direction to model big and varied data. Unlike humans, machine learning improves when working with growing data sets. The more data is fed into an ML system, the more the system will learn, resulting in the production of higher-quality insights.”
Along with an interesting discussion of the “5 Vs of big data” (volume, velocity, variety, veracity, and value), the authors highlight a few interesting cases where cutting-edge companies are using machine learning to make a big difference. Two of these are applications at EasyJet and at PayPal.
EasyJet has been using an ML system to predict what kind and how much food will be needed on a flight, optimizing food inventory on its planes. The system looks at a series of features, such as weather, time of year, type of customer, etc. to make its predictions. The insight is extremely valuable: the company realizes cost savings by avoiding unnecessary expense for food that would be wasted. In addition, when less food is loaded the plane is lighter, resulting in less fuel consumption/expense.
PayPal’s fraud rate is around 0.32% — phenomenally low, compared to other merchants’ 1.32% average rate. How do they do it? With a fraud detection system that uses machine learning. With algorithms that mine data from the customer’s purchasing history — and also review stored patterns of likely fraud — PayPal can determine whether a suspect transaction was, for example, the innocent action of someone traveling for work or if it’s a possible fraud attempt.
Is your organization using, or looking at machine learning techniques to increase the value of all that data it’s collecting?
For More on Machine Learning
Cutter Research: Cutter Consortium clients can read Why Machine Learning Is Crucial to Effective Utilization of Big Data, to learn more about the challenges in big data applications, the 5 Vs of big data, and the machine learning advantage.
Examine different machine learning mechanisms in Refining Reconciliation: A Machine Learning Approach to the Financial Industry’s Toughest Task, where the authors, Zhuo Li and Jianling Sun, propose an approach to fitting massive data sets in the real world of reconciliation and provide a balanced solution to address the high skewness in reconciliation data sets.
And, for a quick primer, read Curt Hall’s Machine Learning Rising.