Microsoft Azure ML Designed to Attract More Machine Learning Development Partners: First Use Case is by OSIsoft at Carnegie Mellon
Microsoft recently demonstrated a special Azure ML cloud service that provides a browser-based interface for authoring the machine learning algorithms that power predictive analytics, such as Smart Building operational analytics. The interface is easy enough for non-data scientists to use, putting the emphasis on users’ ability to insert their own domain-specific expertise into the task of algorithm design. The Azure cloud service is popular among both independent and ESCO-owned Energy Efficiency Management software providers, so the Machine Learning offering should be a boon to the Smart Building industry. According to an article posted at the Redmond Channel Partner site, Azure ML was used by OSIsoft to build the well-publicized Carnegie Mellon University Real-time Building Performance Optimization Solution.
At CMU, OSIsoft leveraged its PI System’s real-time data collection, analytic and visualization capabilities to integrate, monitor and diagnose building performance indices. As detailed in this presentation, Multiple building automation systems were integrated for control through this common platform.
As Allan McHale of research company Memoori explains in this article about the BEMS 2013-2017 market, the big ESCOs can meet most project requirements with BEMS and Energy Efficiency Management (EEM) solutions from companies they own, but they don’t own public cloud platforms. So one of the dynamics in play right now inside the Smart Buildings space is market pressure for traditional Building Energy Management Systems (BEMS) companies to forge partnerships with cloud service providers. Before they sign a contract for a data-driven, continuous optimization approach to building operation, building owners want both their BEMS and IT partners to have worked out any major interoperability issues. OSIsoft and Schneider Electric entered into such an agreement in early 2013.
The Microsoft spokesperson acknowledges that it’s up to partners to make Azure ML a success by developing their own IP around the service, as OSI has done. Microsoft plans to release a software development kit (SDK) when Azure ML hits general availability.