Microsoft versus Amazon: Microsoft has the edge in machine learning, argues Forrester's Mike Gualtieri
Microsoft's lead in machine learning encourages Amazon Web Services to rush out its own equivalent
Microsoft's Azure Machine Learning has the edge over Amazon Web Service's recently launched alternative, according to Forrester Research analyst Mike Gualtieri. However, Amazon's policy of releasing products early and adding capabilities as and when, means that AWS could catch up quickly.
Speaking to Computing, Gualtieri said: "Microsoft's offering is a lot more sophisticated because underpinning Microsoft's solution are two key things that differentiate it from Amazon's: one is the general-purpose analytic tool. It has a very visual tool for creating arbitrary analytic workloads. And, compared to Amazon, it has some light data preparation tools, but they only have one class of linear modelling algorithms."
Underlying Microsoft's sophistication, he continued, is R, an open-source language designed to aid statistical programming. However, the use of R presents its own challenges for enterprise users, warned Gualtieri.
"One of the downsides of picking R is that it was developed 15 years ago. It was designed to run on single-core desktop workstations belonging to scientists and it's not inherently scalable," he said.
Microsoft purchased a start-up called Revolution Analytics, which was formed just two years ago with the mission of making R more scalable by parallelising many of the algorithms in R. However, it will be some time before the full fruits of that acquisition feed through into Azure Machine Learning and, in the meantime, AWS ought to be more scalable.
Both companies also bring their own predictive analytics competencies to bear on the market, he added, Microsoft via its Bing search engine, and Amazon in terms of recommendation engines, based on the technology it uses on its own popular e-commerce website, whereby customers' past purchases are analysed in order to push products they might be interested in.
However, Amazon is very much playing catch-up to Microsoft Azure's machine learning launch last year.
"Amazon released its tool early, but it always releases its services early," said Gualtieri.
"It's strategy is release early - it doesn't need to be the most sophisticated tool in the box - and then they will continue to update it regularly, especially as it gets feedback from customers."