July 16, 2018

Periscope Data integrates with Amazon’s SageMaker to simplify machine learning

By 
Mike Wheatley

Data analytics company Periscope Data Inc. is working with Amazon Web Services Inc. on a new initiative to make it easier to create machine learning models for business decision-making.

Under the initiative, Periscope Data is integrating its data analytics platform with Amazon SageMaker, which is a managed machine learning service that’s used to build and train ML models, and then directly deploy them into a production-ready hosted environment.

Periscope Data said current methods for creating machine learning models are fragmented and time-consuming because they involve using many disparate tools for important tasks such as data preparation and feature engineering. That makes it challenging to access and combine different data sources, which is a necessary step to build more accurate machine learning models, the company said.

To simplify things, Periscope Data wants to combine the process of building machine learning models into a single workflow. The first step involves using Periscope Data’s Unified Data Platform to prepare the training data for these models. The company’s platform makes this easier, since it can surface distribution and descriptive statistics that can help users to understand when more data cleaning is required. The platform also makes it easier for users to iterate on feature selection for each model, the company said.

Once the data sets are ready, they’re then fed directly into Amazon SageMaker, which quickly creates and trains the algorithm ready for deployment. The machine learning model is then deployed to a data warehouse, from where Periscope Data can quickly access the fruits of its labor and deliver insights to users. The company provided this graphic to illustrate the process:

periscope-data-sagemaker

The integration with Amazon SageMaker comes after Periscope Data added data warehousing capabilities to its platform in November, helping its users deal with data fragmentation issues.

“Building trustworthy machine learning algorithms is a huge challenge for today’s data teams, in part because of the difficulties they face with data prep and feature engineering,” said Harry Glaser, co-founder and chief executive officer at Periscope Data.

The new integration with Periscope Data and Amazon SageMaker is now available in beta and will be demonstrated during a joint July 18 event at the AWS Summit New York 2018 conference.

Original Post

Latest News

More News