Machine learning has come to Microsoft's cloud. On Monday, the tech giant announced the launch of Azure Machine Learning.
Machine learning uses historical data to create a model that can be used to predict future events. The new Azure-based offering is a fully managed cloud service for constructing predictive analytics.
"In mere hours, with Azure ML, customers and partners can build data-driven applications to predict, forecast and change future outcomes -- a process that previously took weeks and months," Corporate Vice President Joseph Sirosh wrote Monday in a post on The Official Microsoft Blog.
The service starts in public preview next month, and is designed to integrate the capabilities of new analytics tools and new software developed for Xbox and Bing. Being in the cloud, it is of course accessible from anywhere and requires no start-up costs.
"Data scientists are in short supply, commercial software licenses can be expensive and popular programming languages for statistical computing have a steep learning curve," Sirosh said. Those obstacles can all be overcome by a cloud-based service, he said.
Tasks using machine learning can employ visual workflows and start-up templates, and developers can use the service to publish APIs and Web services quickly. Azure ML also supports more than 300 packages that utilize the open-source R Project for statistical computing. Microsoft is expected to add support for the Python language, which is popular among data programmers.
As a case example, Microsoft cites Microsoft partner MAX451, which used the ML service during its private preview to help an unnamed large retailer decide which products its customers were most likely to buy next, using data from physical stores and e-commerce purchases. This information allowed the retailer to pre-stock its inventory with those products.
OSISoft used Azure ML in conjunction with Carnegie Mellon University to detect faults in real time around campus buildings, as well as to determine various energy output scenaria.
Other applications, Microsoft said, include forecasting demand, estimating disease outbreaks, detecting fraud, predicting crime -- essentially, anything which might benefit from analyzing data to determine a pattern, and then applying that pattern to future events. Cloud access is intended to remove the start-up barriers that companies would otherwise face, as well as making it continuously available from anyplace.
According to ZDNet, the Azure ML service was previously known as the CloudML project, and about 100 customers and partners used it during the private preview phase. Microsoft Stores reportedly use Azure ML to help pinpoint fraudulent transactions, which has resulted in a fraud reduction of 15 percent to 20 percent. The new service includes a design studio for business analysts, an API, and an SDK so that developers can use the service in their applications.
Azure ML comes on the scene following the launch last year of IBM's Watson service, based on the famous supercomputer system that won the TV game Jeopardy and which is now available in the cloud.