Azure ML: Machines are Learning Today!
In my last blog I wrote about Azure ML as a disruptive technology that will bring Predictive Analytics to the masses. Although I have been explaining the concept in mostly theoretical terms, this technology is already being used today, and here are a few examples.
One application of machine learning we have all experienced already is the automated recommendation. When a web site such as Amazon gives us book recommendations based on our previous purchases or Netflix tells us that we might enjoy a certain movie because of our viewing history, that feat is performed by a computer algorithm (an algorithm is basically a program that uses a set of processes or rules to solve a puzzle or problem). This “recommendation algorithm” is constantly analyzing huge amounts of buyer data, including what and when items were purchased by every customer and how they were rated, to “learn” much more rapidly than humans possibly could. And web sites that use these algorithms are constantly modifying them to make them more effective.
A second very valuable application of machine learning is price optimization. Too often what is billed to customers is dependent upon static price lists or on the subjective opinions of individual salespeople or sales managers. Machine learning can utilize factors such as supply and demand, macroeconomic data, seasonality, industry standards and a great deal more to recommend pricing per customer and per item. Obviously if you can optimize what you charge in every situation you will be generating significantly more revenue—and that is an area where we should all invest.
As a final illustration, I refer to Microsoft’s Convergence 2015 conference in March, where a company called Wash Laundry, an outsourced provider of washing machines to residential locations such as apartment buildings and dorms, showcased its use of Azure ML. During its presentation the CIO explained that they learned through their data analysis that the success of their facilities was based on such factors as the local unemployment rate, median household income, and even gas prices. When several months ago they decided to open up in a new city, they used Azure ML to analyze these and other relevant data points to examine multiple cities around the US—and their model determined that Miami should be the next location in which to expand.
So Azure ML is making companies productive today, and in my next blog I will discuss in more detail the capabilities one must have to develop on Azure ML effectively.