Advances in Data Science Applications Presented by Analysis Group Experts

March 19, 2019

Advances in data science are transforming the practice of economic, litigation, and health care consulting. On February 27, 2019, Analysis Group held a Data Science Symposium, which was attended by over 250 consultants and professional staff. The Symposium serves as a forum for the Data Science team and the consultants involved in data science projects to share with their colleagues across the company examples of the innovative methodologies and tools developed in this field.

Four of the sixteen examples presented are summarized below.

  • A case team developed a natural language processing (NLP) tool for a class action lawsuit, capable of parsing a large number of depositions and converting them into a centralized database. The team also wrote machine learning (ML) algorithms to identify the key economic questions to which each deposition was relevant. Analysis Group has built and deployed similar tools across other kinds of unstructured text, such as call or chatroom transcripts, and emails.
  • Our Health Care teams are often asked to estimate the prevalence of a disease in a particular population or sub-population, and to predict patient outcomes for that disease. In a recent engagement, a team was asked to do both for a particularly difficult-to-diagnose autoimmune disorder. Using claims data and electronic medical records (EMRs), the team employed both benchmark and cutting-edge ML algorithms to estimate the share of the population with the disorder, and to predict the likelihood that a particular patient would contract it, based on a number of factors.
  • The ability to parse huge amounts of text to extract relevant information is one of the key achievements of data science. In one case, a client asked Analysis Group for assistance working with algorithmic trading logs. These logs amounted to some 70 TB of unstructured data, containing billions of lines of text recording algorithmic activity at the microsecond level. From this massive dataset, our client needed us to extract, process, aggregate, and analyze specific types of information. Analysis Group built a customized data processing pipeline that could identify, parse, and analyze the relevant information.
  • While tools for analyzing and processing data are of great value, so is the ability to present them to clients in a compelling, easy-to-use format. One area in which this capacity is especially relevant is merger analyses, in which the question of the extent of competition within a particular geospatial region is key. As part of a merger filing, competition authorities often require numerous maps to be submitted, showing the relevant customer catchment areas for merging parties and the locations of all competitors. To make this process more efficient and effective, our data science team has built a fully interactive tool that allows for the rapid production of maps that highlight criteria that may be of interest to a competition authority. Using this tool, our consultants have been able to drastically reduce the time required to produce these maps and, using the interactive interface, have been able to provide clients with real-time insights on potential areas of competitive concern.

These tools, and others like them, have allowed us to extract deeper insights from datasets that in many cases would otherwise be impossible to manage, and also to deliver more refined insights more efficiently, and in user-friendly ways.