Econometrica’s Data Scientists Use AI to Customize Research

As a business management consultant firm, we understand the power of data science to evaluate and inform by collecting and examining information across a multitude of analytical platforms. Modern data science enables our researchers to dig deeper and uncover critical truths in a wide variety of projects for the government and private sectors.

We recently added narrow-based language models of artificial intelligence (AI) to Econometrica’s technical capabilities. AI allows us to bring a fresh perspective and highly customized results to our clients and partners in health, housing, communications, finance, homeland security, maritime and water resources, energy, and grants management.

Two key benefits differentiate narrow-based language AI from its predecessors:

- Upgraded search capabilities: Our data scientists can search and organize large amounts of data more quickly and efficiently.
- Highly detailed customization: The AI learns concepts based on words with similar uses, including synonyms, abbreviations, and common misspellings to produce more robust search results.

“Machine learning is newer in the areas we typically work. There are not many people in the world who know how to build narrow-based language models,” says Douglas Schuweiler, a data scientist in Econometrica’s Data Analytics Group. A specialist in behavioral neuroscience, Dr. Schuweiler studied how the brain learns by building artificial neural network models while earning his Ph.D. at Illinois State University.

“Using this newer level of artificial intelligence, we can build a model that is tailored to each client and focused on the data that needs to be analyzed,” says Dr. Schuweiler.

Data Science in Action: Program Evaluation

Recently, Econometrica was tasked by the Federal Emergency Management Agency (FEMA) with evaluating a nationwide grant program for disaster preparedness. In other words, were the projects generating the intended outcomes? We had to determine the effectiveness of 100,000 individual grant projects, but there was a great deal of latitude in the outcomes each project could achieve.

We used narrow-based language AI to analyze the project descriptions written by grant recipients. “It allowed us to form a more precise evaluation by matching the grant programs with their targeted outcomes,” says Dr. Schuweiler.

This summary of the FEMA project is simplified, but completing this task takes a significant time commitment from specialists with high-level expertise. Before comparing grant project descriptions to their intended outcomes, our team had to develop another model to differentiate project descriptions with a high level of detail from project descriptions with a low level of detail. The process took more than 6 months.

Data science is the linchpin in many of Econometrica’s projects. We use data analytics in statistical sampling, machine learning, data scrubbing, economic modeling, and analysis support for audits, regulatory work, improper payments, and predictive analytics. This support includes the following:

- Development of detailed sampling plans.
- Advanced statistical analysis of the results.
- Comprehensive reports that document the methodologies used and certify the statistical validity of the analysis.

One recent data science project earned an Award for Excellence for our Data Analytics Group, based on the real-world impact it created. You can read about it here.

Work With Us, Work for Us

Econometrica specializes in data analytics for the public and private sector. Our team represents the best and brightest in data science, health, grants management, energy, homeland security, housing and community development, capital markets and finance, and transportation. We work as the lead service provider, and also as an outsource partner to other consultancies. For exceptional results on your next project, visit us online and email a member of our executive staff in your preferred specialty. To explore the benefits of working for us, visit our careers page.

ECONOMETRICA BLOG