A Collaborative Approach to Data Science

Big Data

Big Data / Big Data 17 Views 0

Organizations have a rapidly expanding amount of data with which to work, but deriving insight from that data is a challenge — not least because of persistent skills shortages in the industry. Data analysis typically begins with the identification of so-called "features," which are data points that are believed to have predictive powers. The identification of these features is usually something that requires a degree of experience.

A team from MIT has developed a new tool called FeatureHub that they believe will make feature identification easier. The tool, which is documented in a recently published paper, takes a collaborative approach to the task, with data scientists working together to review a problem and propose features that are then tested by the software against target data to gauge their usefulness.