Machine learning is changing the analytics picture at CoreLogic Inc. -- literally.
CoreLogic provides information on real estate, mortgages and consumer credit to lenders, insurers and government agencies. The info it collects includes "millions and millions and millions" of photos, says Robin Gordon, the Irvine, Calif., company's chief data officer. And now, it's using machine learning technology to analyze the images and generate additional data about properties.
"We can start to extract data we could never extract before," Gordon says. For example, she points to information about roof conditions that can be pulled out of aerial photos taken by drones, giving customers "a much richer view" of buildings for weighing mortgage and insurance applications. But it's not just about photos. Gordon says CoreLogic is also using machine learning software to extract data from forms filed with county governments "instead of having an army of people offshore keying it in."
These days, CoreLogic has plenty of company in tapping machine learning tools. After being a niche technology for decades, machine learning is stepping more into the analytics mainstream along with other forms of artificial intelligence, such as deep learning and cognitive computing.
But machine learning applications often require new analytics systems and tools. They also create new data management challenges, partly due to the large amount of data that's typically involved.
In addition, organizations need to find machine learning uses "that will have a meaningful impact" on business operations, TDWI analyst Fern Halper wrote in an April 2017 report. Initially, that was an issue for Cisco Systems' corporate data science team. In early machine learning projects, "we weren't answering or solving the right questions," senior data scientist Anu Miller says.
This handbook offers insight and advice to help make sure your organization's use of machine learning technology and techniques is a pretty picture, with a sharp focus on business benefits.