BT

Facilitating the Spread of Knowledge and Innovation in Professional Software Development

Write for InfoQ

Topics

Choose your language

InfoQ Homepage News Microsoft Expands Azure Machine Learning and Real Time Analytics Offering

Microsoft Expands Azure Machine Learning and Real Time Analytics Offering

Microsoft recently announced new machine learning capabilities for Microsoft Azure platform. Azure marketplace now has more machine learning capabilities as web services. A vast array of services from Microsoft and third party providers provide end users with solutions from anomaly detection to regression models, binary classifiers, forecasting and others. Examples of web services offered off the shelf are a recommendation engine for adding product recommendations to a website and a fraud detection system.

Developers can also create their own web services and publish them to Azure Marketplace. Using the Azure ML Studio GUI, one can create the custom endpoints, get the custom code provided in C#, Python or R and deploy her web service for free or by defining a tiered paid model.

Microsoft also announced availability of Apache Storm for Azure. Storm is a distributed realtime computation system. Storm can be used with HDInsight, Microsoft’s distribution of Hadoop for Azure, to build real time analytics. Like Amazon AWS, now Microsoft Azure also supports Storm for real time analytics. Both announcements were made at the Strata + Hadoop World conference in New York.

Azure Stream Analytics, a new offering announced at TechEd Europe 2014 can be deployed alongside Storm for data transformation purposes. Azure Stream analytics is also architected for large scale, real time event processing and allows developers to use a SQL-like syntax that can speed up development. In addition, it integrates out of the box with Event Hubs, the newest event queueing system for Azure. Event Hubs can be deployed as a drop in replacement for Kafka or ActiveMQ systems and acts as a time-based buffered broker for messages.

Together with Data Factory which has just been announced, developers can now ingest data from heterogeneous data sources with both structured and unstructured data. Data from on-premises and cloud sources can be combined and form data pipelines that can be monitored using Azure powered tools.

Microsoft Virtual Academy organized a live event this week, hosted by Seayoung Rhee and Buck Woody. In this almost four hours long webinar they went all the way from introduction of the ML Studio to designing a sample predictive analytics solution and publishing it in the Azure marketplace. Microsoft Azure Machine Learning as a Service has gone a long way since its launch and it sure looks like Microsoft is investing on its long term success.

Rate this Article

Adoption
Style

BT