Explorium advises companies on improving and accelerating analytical machine learning

The key to successful artificial intelligence-based advanced analytics training is augmenting internal data with external data, according to data science platform startup Explorium Inc.

The process is not easy, however. The problems include obtaining and managing this mass of external material.

“The key to any analytical problem is having the right data,” said Zach booth (photo), Director of Global Partnerships and Channels at Explorium.

Models are as strong as the data they train on, but getting the external data you need is difficult. “It’s manual, it’s tedious and it’s extremely time-consuming,” he said.

Explorium claims to have a solution that includes an organized catalog of data sources, coupled with a platform to integrate everything into an organization’s existing internal hardware.

Booth spoke with Jean Furrier, host of theCUBE, the live streaming studio of SiliconANGLE Media, at the AWS Startup Showcase: The Next Big Things in AI, Security, and Life Sciences. They discussed how external data source platforms – leveraging the massive amounts of data generated globally – could improve and accelerate analytical machine learning for individual organizations. (* Disclosure below.)

Context of data needs for analysis

Companies need to bring variety to analytical pipelines, according to Booth. The way it is done is to add more data, and in particular to augment the existing and high quality data produced by the organization with external data.

Internal data is good data, Booth cautions, but the problem is that the data created by the organization lacks context. Raising it with a wide variety of additional data connections in the cloud solves this isolation problem. “We help clients access better and more relevant external data to fuel their predictive and analytical models,” he said.

Fraud detection is one use case that will benefit, according to Booth. Verticals could include financial services, insurance, e-commerce, consumer goods and software as a service.

Decision making to generate and create actions through traditional analog means, such as rule-based approaches or using simple spreadsheets, is just not dynamic or flexible enough. “It’s very limited in its ability to change,” Booth said. So you have to plug in analyzes and do the work via the cloud.

Explorium is part of the network of Amazon partners with the aim of providing context to decision-making.

“Modeling and using data really is a huge arsenal at your fingertips,” Booth said. “The trick is to extract value.” Filtering, navigating and connecting to the increasing amounts of data created outside of an organization is the way he wants to do it.

In the case of fraud analysis, the ultimate goal of reducing fraud is to break down how external data can be integrated into three steps, according to Booth. The first is to take advantage of the fact that businesses have historical data showing how many customers have acted fraudulently over time. This is the internal training data. It can be provided to Explorium as an Excel spreadsheet, for example. This initial data produces a binary output indicating whether the business was fraudulent or not.

Data enrichments are then added in a second step by the machine learning analytical platform. The purpose of this step is to generate signals, called “characteristics” in machine learning.

The platform then matches and connects internal and external data, inferring and understanding the meaning of the data while automating the process of signal distillation.

This three-step process generates “an advantage in your modeling efforts,” said Booth. Fraud rates can thus be reduced and customers can benefit from better prices or other benefits, such as more convincing service.

Organizational rationalization

But it is not only the offers of end customers that benefit. An organization itself also sees great value, says Booth. This is because data scientists can spend more time engineering features and signals and less time managing and formatting data. One of Explorium’s platforms, called Signal Studio, is for data science teams looking for external data signals for analysis pipelines.

“You don’t have to go out and put all that one-time effort into finding data, organizing it, cleaning it up, and so on,” Booth said. Large-scale modeling is also made possible by the platform.

Interestingly, assessing whether there is a return on investment is another area that Booth says can be optimized by using an external data source platform for analytics data. The time spent manually searching, cleaning, validating, etc. harms ROI if the model doesn’t work, which takes time. It becomes an opportunity cost.

“It takes months,” Booth said of the manual method. “Every day more and more data is created outside of our organization. Can I use a tool that can efficiently query all of this data? Is the question organizations should be asking themselves, he said.

Stay tuned for the full video interview, which is part of SiliconANGLE and theCUBE coverage of the AWS Startup Showcase: The Next Big Things in AI, Security, and Life Sciences. (* Disclosure: Explorium sponsored this segment of CUBE. Neither Explorium nor other sponsors have editorial control over the content of CUBE or SiliconANGLE.)

Photo: SiliconANGLE