MIT built a new DATA SCIENCE MACHINE able to translate human intuition for Big-Data Analysis

Big-data analysis involves searching for suppressed patterns that possess some form of predictive power. But selecting which “features” of the data to examine normally requires some human intuition. In a database encompassing, say, the beginning and end dates of numerous sales promotions and weekly profits, the vital information may not be the dates themselves but the areas between them, or not the entire profits but the averages amid those spans. MIT researchers are targeting to take off the human element out of big-data analysis, with an innovative method that not only hunts for patterns but designs the feature set, too. To examine the first prototype of their system, they registered it in three data science competitions, where they contested against human teams to detect predictive patterns in unaccustomed arrays of…


Link to Full Article: MIT built a new DATA SCIENCE MACHINE able to translate human intuition for Big-Data Analysis