SDAD Projects

Publications

• Designating Skilled Technical Workforce Using Online Job Postings (2024)

with Lancaster V, Montalvo C. (Conferences: JSM, FCSM)

• Detecting Pandemic Related R&D Trends Using Dynamic Topic Modeling (2023)

with Linehan K, Thruston J, Shipp S, Kindlon A, Jankowski, J.

• Detecting Federally Funded Research and Development Tends using Machine Learning and Information Retrievals Methods (2022).

with Linehan K, Oh E, Thruston J, Garrett M, Keller S, Shipp S, Kindlon A, Jankowski, J.

• Women Inventors in the US Patent System an Exploratory Analysis and Literature Review - a Pilot Study (2022) 

with Linehan K, Kattampallil T, Shipp S.

• Designating the Skilled Technical Workforce Using O*NET-SOC (2021) 

with Lancaster V, Montalvo C.

Ongoing Projects

• The Broadband Initiative Program Impact on New Construction (forthcoming) 

with Golstein J, Kattampallil T.

This paper investigates the impact of the Broadband Initiative Program (BIP) on new construction in rural areas. Using property sales data from Corelogic, we estimate the count of new construction at the census tract level. The BIP's impact is assessed by comparing the difference in new construction between tracts within the project area and those outside the project area (within a 25-mile buffer from the program area). Tracts inside and outside the project area are matched based on similar characteristics. Using a difference-in-differences model combined with a zero-inflated Poisson model, our analysis reveals no significant impact from the BIP on new construction in rural areas.

• Model-based Approach of Identifying Minority-owned Business: Case of Fairfax County (forthcoming) 

with Treena G.

This paper proposes a model-based approach to identify minority-owned businesses in Fairfax County. Leveraging microdata on companies from Mergent Intellect, we construct a classifier for minority-owned businesses using a dataset from Mergent Intellect. The training sample was generated by classifying companies as minority-owned or not through various sources, including scraping data from platforms such as Yelp and Chamber of Commerce, along with references to the Civil Rights Act of 1964. To predict the probability of a company executive being non-white based on their names, we use pre-trained models like raceBERT and ethnicolor. Our classifier incorporates features such as the average predicted probability of executives being non-white, the percentage of non-white residents at the business location, and the company name. Our model predicts that 45% of companies in Fairfax County are owned by minorities, a figure closely aligned with the 37% reported by the Census.

• Data Commons (see the project details)

  • • Mastercard & Fairfax County: - Estimate demographics at the small geographic areas (parcel level)  - Explore the business diversity in Fairfax County.

    • Virginia Department of Health: Build the Health Opportunity Index (HOI).