First Analysis Group Data Science Symposium of 2021 Updates Firm on Recent Applications of Advanced Analytics and Machine Learning Tools

June 9, 2021

Using modern data science techniques, Analysis Group’s case teams solve real-world problems with more speed and efficiency than ever. In May 2021, several data scientists and collaborating consultants hosted the first of three Data Science Symposiums of the year to update the firm’s staff on the latest tools and innovations applicable to case work. This installment covered several examples of advanced analytics and their application.

  • With the COVID-19 pandemic bearing down on Haiti, Analysis Group was asked to help forecast regional outbreaks so the country could swiftly and effectively distribute a forthcoming shipment of critical medical supplies to the regions where they would be needed most. A lack of Haiti-specific data about the spread of the disease created data gaps that didn’t allow for traditional epidemiological forecasting models. In just a few weeks’ time, the case team leveraged machine learning (ML) to develop projections based on other, comparable countries’ use of lockdowns at various points in their COVID-19 outbreaks and data on the related public health impacts. This deep-learning model provided accurate predictions to Haiti’s Ministry of Health about the country’s key outbreak patterns. Such predictive capabilities are especially valuable in a variety of health care cases where continual, real-time model predictions could help advance new treatments and save lives.
  • Clients benefit from the use of optimization algorithms in many different applications. One Analysis Group team used such an algorithm to develop an impact assessment of new electricity market rules on energy security. In creating a model for simulating future-world conditions, the team considered approximately 40,000 variables defining the entire power grid’s activity over each hour of the day, as well as approximately 40,000 constraints that represented current and new products over each hour. Another case team used optimization in a merger between two major supermarket chains to analyze approximately 5,000 markets and identify the optimal set of divestitures that would meet the regulator’s conditions for approval while maintaining the merged company’s key business objectives.
  • The ability to match records across two or more datasets that refer to the same entity is central to many analyses across multiple practice areas. Analysis Group has developed tools and processes for addressing this need, and has combined them in a state-of-the-art customizable app that can be applied to a variety of matching tasks. Combining numerous matching algorithms with coordinated parallel review pipelines, the app allows teams to narrowly and efficiently focus their manual review on matches that require human judgment, and to review all relevant information and confirm matches with a few clicks. This highly adaptable matching interface already has many applications, including in health care, technology, finance, and pro bono cases that require matching capabilities when combining massive or numerous datasets.

At the remaining symposiums, the firm’s data scientists will discuss methods for working with large amounts of text, as well as interactive, web-based economic models.