Follow Datanami:
July 24, 2017

DataRobot Reaches Out to SAS, Financial Services

Companies that use DataRobot’s software to automate data science tasks can now output models directly from SAS, the dominant analytics company whose software is widely deployed in enterprises around the world. The upstart analytics automation firm also unveiled updates that specifically target the financial services industry.

DataRobot is one of a number of software companies seeking to make production machine learning a lighter lift, and enable moderately skilled business analysts to participate in data science projects using the latest open source technologies, including Apache Spark.

The company seeks to accelerate data science work by eliminating repetitive tasks, Razi Raziuddin, DataRobot’s vice president of strategic business development, told Datanami earlier this year.

“The real challenge is fitting the algorithms, shaping the data, and doing all the data prep and transformations that are needed in order to feed the data into the algorithms, and then tuning the algorithms to build a really good model that works really well for a given data set,” Raziuddin said.

“That is a big manual task right now.  It requires PhD data scientist who have grown up doing machine learning or doing predictive analytics for a number of years who also understand the domains and the nuances of data in a particular domains. What we’re trying to do is take a lot of the repetitive task and a lot of the science, if you will, and automate it.”

With the latest release unveiled today, DataRobot is reaching out to SAS, the North Carolina firm that dominated the field of analytics for decades. While SAS does integrate with open source products like Hadoop and Spark, its own software remains proprietary, which puts it at odds with the mainstream of data science today and its focus on open source.

The DataRobot software can now ingest SAS files directly, which the company says will lower the barriers for DataRobot customers to work with SAS or even replace it. “This will smooth the transition for many organizations as they look to move beyond manual, time-consuming modeling approaches, and will de-risk their adoption of open source technologies,” the company says.

Today’s release also brings new features for the financial services industry, specifically a new type of model called a Generalized Additive Model that the company claims will help insurance companies make better pricing decisions with their own products.

“These newer sets of algorithms will allow users to solve pricing and risk segmentation problems more accurately, without sacrificing transparency or interpretability,” the company says.

Last but not least, the Boston, Massachusetts company unveiled a new feature that allows customers to generate scoring code in Java. Previously, the scoring mechanism (which allows models to be actually put into production) was limited to running via DataRobot’s API or deployed via Spark. Now that it’s generating Java, the code can run practically anywhere, the company says.

DataRobot CEO Jeremy Achin says there are many more enhancements in the company’s pipeline. “Our next few releases are going to shake the worlds of business, data science and beyond,” he stated in a press release announcing the new features.

Specifically, the company has plans to integrate the capabilities it obtained with its May acquisition of Nutonian, an analytics firm that specialized in modeling time-series data. DataRobot says the new release of its software contains architectural changes that will enable Nutonian’s capabilities to be embedded into the DataRobot platform.

That is expected to occur later this year, the company says.

Related Items:

Taking the Data Scientist Out of Data Science

DataRobot Delivers an ML Automation Boost for Evariant

Real Machine Intelligence Making its Way into the Enterprise