Automated Machine Learning
A shift away from presumptive analytics
Conventional analytics require analysts to first form a hypothesis — a question — and then query the data to surface the answer to that question. In order to arrive at accurate results, this model presumes that 1) the analyst knows the right questions to ask and 2) the hypothesis and the resulting insights are free of bias. But, of course, both of these parameters are impossible to achieve: Humans can’t possibly know all the right questions and, by our very nature, those questions are loaded with bias, influenced by our presumptions, selections and what we intuitively expect to see.
In 2016, we’ll see a strong shift from presumptive analytics — where we rely on human analysts to ask the right, bias-free questions — toward automated machine learning and smart pattern discovery techniques that objectively ask every question, eliminating bias and overcoming limitations.