Microsoft wants to become a leader in machine learning and has been spruiking the wonders of this technology in helping businesses make better decisions. That’s good and all, but does Microsoft actually use it internally in its own organisation? Apparently so, and it has saved the company millions.
Roboting arm picture from Shutterstock
Machine learning involves developing algorithms for computers to study patterns in data to make predictions. It’s an exciting field of computer science in a world where data is being generated at an alarming rate. A number of industries including retail, farming and even graphics design have embraced the technology. If you’re still stumped on what machine learning is, we have an article to help you better understand it in a single PowerPoint slide.
One company that is heavily invested in machine learning is Microsoft. It uses it in every one of its products, including for the Bing search engine and, more recently, Cortana who is an online personal assistant that has been integrated into the Windows 10 operating system. While it’s nice to have Cortana to schedule in appointments and search for the best restaurant to go to, she can do so much more, especially within an enterprise.
At a recent Microsoft media event, the company showed off a video depicting the potential applications of Cortana for business.
Picture a worker in a healthcare organisation who wants to check in on the status of a customer and being able to ask Cortana to do just that. The program then pulls data on that customer from every part of the organisation to come up with their current status and predict their health condition in the next month based on that information. Microsoft sees this as the future for the enterprise sector.
Practice What You Preach
Putting money where its mouth is, the company incorporates this predictive analytics technology into its own operations. Microsoft vice-president for machine learning, Joseph Sirosh, told us that machine learning techniques have saved the business millions of dollars every month.
“We use it for fraud detection. Every transaction placed on Microsoft’s website are screened by machine learning models behind the scenes. It has saving us millions of dollars every month because of its power to detect what’s a good and bad transaction,” he said. Another way Microsoft uses this predictive analytics method is for sales forecasting.
“We use it to forecast quite accurately what our sales are going to be in the future and it is actually much more accurate than what our own financial analysts can come up with,” Sirosh added. This is an area where he believes these predictive analytics tools will thrive in the enterprise.
“Every finance department of every company is constantly generating forecast using Excel spreadsheets and their own calculations. What if you could use a lot of different types of data and machine learning to predict it in a data driven way? What the output will be and what the actual sales are going to be. you can actually come up with much better numbers, which is why we are using that internally,” Sirosh said.
Microsoft has stiff competition in this space. IBM has its Watson, which has garnered a lot of attention as it is applied to a wide variety of tasks such as identifying cancer and, most recently, helping people win at Fantasy Football. This year, Amazon Web Services came out with its machine learning-as-a-service platform.
All this development in the realm of commericialising machine learning is making this technology more accessible to organisations that may not have the resources to hire their own data scientists to sieve through vaults of information and distil it in a meaningful way for their businesses. As more vendors wrestle for dominance in this area, the costs and barriers to accessing this technology will come down making it easy to use machine learning even for smaller companies or one-man band businesses.