Automated machine learning company DataRobot raises $54m Posted on : Mar 26 - 2017

DataRobot has raised $54 million in the first close of a Series C round led by New Enterprise Associates.

The latest round brings the total amount raised by the Boston, Massachusetts-based company to $111 million, with "significant" additional funding expected in the second close of the round.

Data scientists Jeremy Achin and Thomas DeGodoy founded DataRobot in 2012 on the belief that automated machine learning will not only increase productivity for data scientists, but will also open up the world of data science to non-data scientists.

The DataRobot platform features hundreds of open-source machine learning algorithms, allowing users to quickly build predictive models.

Chris Devaney, COO at DataRobot, told that a data scientist would typically look at a set of data, prepare that data, and then train a predictive model -- a process that can take weeks or even months.

"In other words, the data scientist would take information they already know the answer to and then train algorithms to be able to predict the future. For example, they would look at fraudulent behaviour of the past to predict fraudulent behaviour in the future," Devaney said.

"Typically, it would take them weeks or months to come up with an algorithm that they feel comfortable with to make predictions on future data."

Whereas with DataRobot, users upload their data, select their target variables, and the platform "automates, trains, and evaluates".

"A data scientist could look at their existing models, enter that into DataRobot along with all the open source models, and see how it ranks. Since it only takes minutes to find the best model, you can run and rerun the competition to find the most accurate model every day or every week, depending on how often your business changes. If a new model is more accurate or becomes available, then you simply point to that new model," Devaney said.

"When you think about fraud, a fraudster will continue what they're doing until they get caught. If an institution takes months to update the algorithm to catch that behaviour, the opportunity loss is very significant. So DataRobot's ability to refresh models and to always find and have the most accurate model is very compelling for large institutions where accuracy can really translate to tens or hundreds of millions of dollars in savings."

To date, DataRobot has created more than 150 million predictive models for a range of customers including sports teams, banks, insurers, and healthcare providers.

DataRobot said it will use the latest investment to double the size of its research and development team to speed up development of new features and capabilities. It currently has about 150 employees worldwide, half of whom are based in the company's Boston headquarters.

The company will also use the funds to continue its global expansion -- it currently has offices in Europe, Japan, and Singapore -- and bring increased support for global channel and technology partners including Cloudera, Alteryx, global systems integrators, analytics solution providers, and resellers. Source