Data-driven Access of Philanthropy
The skills the authors demonstrated here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Does data visualization provide any shortcuts for aligning “doers” with “funders”?
Philanthropy has been funding change-makers to improve the lives of countless people, stimulate innovation, and foster healing relationships within communities across the globe for centuries according to data collected.
Despite the phenomenal wealth amassed by private foundations, applying for these funds is highly competitive, demanding dedicated resources from already “stretched” non-profit organizations. Even more discouraging, many of these foundations accept applications by invitation only.
Like the for-profit world, decisions to invest are established and fostered through trusted relationships. While social media enables anyone “access”, it doesn’t substitute for the assurance of mutually-aligned goals or values that comes from meaningful dialog and time invested.
So what, if any, value do large datasets of past transactions provide the nonprofit? I was curious to find out.
The Dataset
The dataset I evaluated was sourced from Foundation Directory Online (FDO), a subscription-based grants database of more than 24 million grant records from over 240,000 private foundations. I created the dataset by exporting all grant transactions from 2015 through October 2021 that include the keyword, “climate”.
It captures over 10,000 records and 11 attributes which include:
- Grantmaker name, most commonly a Foundation name
- Grantmaker state
- Recipient name, the non-profit organization or university awarded the grant
- Recipient city, state, and country
- Primary subject, categorizing the grant’s primary focus
- Year in which the grant was authorized
- Support strategies, describing how the award will be used (capital building, general program, unrestricted, etc.)
- Description, describing the program being funded; varies greatly in length and in some instances is absent
Observations of data
Unlike my previous spreadsheet tool of choice, Python pandas enabled me to manipulate the dataset quickly by filtering, reshaping, and aggregating. Matplotlib was my plotting tool of choice.
One meaningful view is histograms reflecting transactions by year as shown below (one caveat: note the scales!). Keep in mind that this dataset is limited to transactions that include the keyword “climate”, and fluctuations are heavily influenced by other current events (e.g., the pandemic and social unrest of 2020).
Reasoning about data
While simple queries and endless pivot tables can help a nonprofit stay current with grantmakers aligned to their operations, the results are not novel nor do they point to a meaningful path in terms of how to connect.
It was at this point that I nearly decided to pack this dataset up when I recognized the significance of the “inner” join: that I could tease out a discrete list of re-granters. This is a nuance not easily wrangled by typical spreadsheet software (or at least, I never set about trying). I was excited to pursue it!
Re-granters are organizations that both receive foundation support as well as distribute their own. Think of these organizations as intermediaries for the larger, exclusive entities with a deeper reach into the communities their recipients serve. In this regard, they are multiple networking steps closer than say, Bill or Melinda Gates, Michael Bloomberg, or any other executive of a major foundation.
With the assistance of NetworkX, I went to work filtering my dataframe with a list of re-granters and plotting the results. I couldn’t have been more excited.
Act!
These types of plots trigger action: who do I need to know?
Clearly, someone associated with a “hub” would be a great place to start!
And while those connections may seem to be too many degrees of separation away, why not try a spoke? I’m excited to think of how these types of visuals can spawn further inquiry (e.g., what are the networks associated with each of the spokes?).
With this in mind, a Python-based NetworkX graph can provide meaningful direction and focus to the non-profit professional charged with securing funds (not to mention maintaining operations, volunteer management, fundraising campaigns, and any other number of unexpected surprises).
Of course, the output this method generates only becomes more meaningful with transaction histories based on specific programs (e.g., reforestation, youth activism, etc.)--I'm looking forward to maturing this workflow for the organizations I am partnering with!
Questions or comments? Let me know!
About Author
