Microsoft Cloud Predictive Analytics Preview Coming Soon
A preview of Azure Machine Learning, or Azure ML (project code name "Passau"), will be made publicly available in July 2014.
June 20, 2014
![Microsoft Azure Machine Learning screenshot Microsoft Azure Machine Learning screenshot](https://eu-images.contentstack.com/v3/assets/blt07f68461ccd75245/blt4816cfaf672b6281/66193ccc4c182e20e8ad89c8/azureml_0_5.jpg?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
Exciting news came out of Redmond this week for advanced analytics with Microsoft Azure Cloud. A preview of Azure Machine Learning, or Azure ML (project code name “Passau”) will be made publicly available in July 2014. Public preview release timing was not announced yet, but it might align with the World Partner Conference (WPC) that will be held on July 13 -17. At the PASS BA Conference held back in April, Microsoft executives mentioned additional big business intelligence (BI) and analytics news would be coming this summer. According to press releases, Azure ML has been in use by approximately 100 customers in the Technology Adoption Program (TAP).
Related: Microsoft Azure Machine Learning Moves Predictive Analytics into the Cloud
A Sneak Peek of the Azure ML Preview
From the early sneak peek information provided and the Official Blog, Azure ML appears to support the full CRISP-DM (Cross Industry Standard Process for Data Mining) lifecycle using Microsoft-developed tools and templates on the Azure cloud as a cloud-based machine-learning service. Azure ML includes a design studio tool that is designed for power user/business analysts versus data scientists, an application programming interface (API) service, and a software development kit (SDK).
From the provided media images, we can probably assume that there will be capabilities to upload a dataset, optionally apply various dataset transformations in a workflow, design an experiment using module building blocks, train the predictive model, evaluate the predictive model, review predictive model scores, and create a Web service with the finalized predictive model that can be called from applications. Additional functionality is unknown at this time.
Cloud Predictive Analytics is Already Widely Used
Despite cloud data security, regulation, and compliance concerns, cloud predictive analytics is already in mainstream use today. It has been offered by many of the top traditional predictive vendors such as IBM, SAS, and FICO, and also a few niche players such as Predixion, BigML, and Alpine Data Labs. According to a 2013 Decision Management Solutions survey of over 350 companies from wide range of industries, close to 90 percent of respondents have at least one cloud predictive analytics solution already deployed. The main reasons cited for using cloud for predictive analytics projects includes greatly reduced costs, massive cloud scalability, and pervasiveness. Typical use cases included pre-packaged, cloud-based decision-making solutions that embed predictive analytics, cloud-based ad-hoc predictive modeling, and predictive model scoring. This is an area of high growth and interest due to the exceptional ROI that can be achieved from proactive, predictive projects.
Predictive Analytics is Not New for Microsoft
Machine learning, data mining, and predictive analytics is an area of statistics around capturing relationships between explanatory variables and predicted variables from past occurrences and using it for predictive purposes. Data mining is not new to Microsoft. It has been available within Microsoft SQL Server Analysis Services (SSAS) for many years now, though few people realize it's available or realize the power it can bring to your business.
Despite the new cloud Azure ML offering announced, it still looks like the older, on premise SSAS Data Mining functionality and Excel Data Mining Add-Ins are being actively promoted on the Microsoft website. There has not been any news of a deprecation that I'm aware of today. In a previous article, Taking SQL Server Analysis Services Data Mining Further, I mentioned caution when using the current on premise data mining offering due to a lack of significant investment. The current on premise solution is fantastic, still functions, and the DMX predictive queries are easy to embed into both applications and "smart reports."
Read more about:
MicrosoftAbout the Author
You May Also Like