Report: The Story Machine Learning Tells About Nonprofits
The analytics firm Civis says that its topic modeling-based analysis of nonprofits highlights how there are different kinds of organizational hot spots around the country.
What could machine learning teach us about nonprofits nationwide? For one thing, a little more about what kinds of nonprofits are more common in different places.
Recently, the data analytics firm Civis teamed with the magazine Fast Company to dive into data about nonprofits in 15 different topic areas. The firm used topic modeling, a type of machine learning, to analyze 2017 tax filings from 353,000 nonprofits secured from the IRS and managed using Amazon Web Services.
The Civis report [registration] found that, in most states, nonprofits that serve vulnerable populations, such as children, seniors, and people with disabilities, are most common, but there are some notable differences in certain regions. For example, three major West Coast states (Arizona, California, and Washington) most commonly focus on youth character development, while key mid-Atlantic states (Maryland and Virginia) are mostly driven by economic development.
There are differences in the funding picture as well—healthcare nonprofits in the Northeast, West Coast, and Central Plains states are generally the best-funded, while labor groups are generally better-funded in the Southeast and the Northern Plains. Meanwhile, benefits and insurance nonprofits tend to get the best funding in the Midwest and a handful of other isolated states around the country (Nevada, New Mexico, Arkansas, New Jersey).
One key lesson highlighted by the report? “The most common type of organization isn’t always the best-funded.”
The report also noted that cities tended to be hot spots for specific kinds of organizations. Youth character development groups were often very common in major cities, while New York City had a strong focus on cultural organizations. But generally healthcare and benefits organizations tend to drive the most revenue despite often not being the most common type of nonprofit in cities.
In comments to Fast Company, Civis Director of Applied Data Science Scarlett Swerdlow notes that the study offers an opportunity to think about nonprofits in a new way. “This is a way for us [to] go deeper and understand the needs that the nonprofit sector is serving and which organizations are receiving the most support and funding to do that work,” she said.
The magazine notes that the report might help nonprofits figure out where to target their efforts to maximize their fundraising and organizational prowess. However, it adds that there are some gaps that might not expose themselves in the current study, as it is only based on a single year of analysis.
(filo/DigitalVision Vectors/Getty Images Plus)