Data Science for the Social sector
Data Science for the Social sector.
By Brian Dalessandro, DataKind Data Ambassador
Jake Porway, DataKind Founder and Executive Director .
At DataKind, we see countless opportunities for data science to make a huge difference in the social sector. Nonprofit organizations are beginning to realize the vast potential of previously untapped datasets for gaining insights about their work, and data scientists are realizing new and exciting ways to do good with data. In order to share what we’re seeing across this burgeoning “data for good” movement, I thought it would be best to hear from someone who’s been on the front lines of our work. Brian Dalessandro is VP of Data Science at Dstillery by day and a DataKind Data Ambassador by night. He has fearlessly led teams of DataDivers through harrowing weekend events to analyze the UN’s survey data on global needs and hunkered down on a DataCorps project to help the NYC Parks Department measure the effectiveness of their tree pruning programs. He has been instrumental in this work and stands as a paragon of the kind of creative, humble, and inspiring data scientists that step up to use their skills for the greater good with DataKind.
Written by Brian Dalessandro
Data is not an end in itself; it’s a means to an end. The collection and storage of data won’t alone change an organization; rather, it is through thoughtful application that data proves its value. Many organizations and industries have been data-driven well before “Big Data” became a trending topic. Within those industries, it’s easy to take for granted that the data-driven culture was created and nurtured through years of trial and error (which some may prefer to euphemistically call A/B testing).
Much of the nonprofit sector is only now forming its own norms around data-driven decision making, with tremendous potential to transform the way these organizations target their services, refine their programs and ultimately scale their work addressing critical social issues. However, many organizations face two major hurdles in adopting more data-driven decision making: a skills gap and a literacy gap. They don’t have access to the rare, expensive data scientists on Wall St. and Silicon Valley that could make sense of their data, and they often don’t understand how they would use them even if they did.This is why data science professionals can make a big impact on the world by sharing their expertise and volunteering their time with these dedicated yet often resource-constrained organizations.
I’ll lead by example on this topic and provide my own experience as a guide to creating a more data-driven culture within nonprofit organizations. I have over a decade of industrial data mining experience, having worked in both incredibly small start-ups as well as Fortune 100 companies. In all of the places I have worked, two necessary components existed to support my success: 1) an existing data infrastructure and 2) the faith and expectation that data will drive value. In my free time over the past few years, I have participated in the efforts of DataKind, a nonprofit dedicated to harnessing the power of data science in the service of humanity, to donate my expertise to organizations in the nonprofit and public spheres. I have learned that they all have the second condition and varying degrees of the first condition above, but most of them lack a person like me.
Data scientists are in high demand, and their salaries reflect that. Most nonprofits and mission-driven organizations simply don’t have the funding to access to this kind of talent, which is why volunteering can be such a powerful resource. My experience working with these organizations has mostly been through my pro bono work with DataKind. For those generous enough to volunteer their time, the key to making any real impact is in your committment to seeing your work through. Engineering a successful data program requires learning, and learning takes time. Fortunately, a few hours a week over a moderate period of time can be enough to make a difference. A pro bono data scientist will likely build new tools and applications for his/her host nonprofit. It is critically important to see these tools through to implementation and ensure the organization has the training and resources to maintain them going forward. Once in production, the investments pay dividends in the scalable reuse of the tools.
Beyond commitment, creativity in applying data science principles to new problems is also essential. For example, the tools data scientists build and deploy in their day jobs might be exactly what a mission-driven organization needs. In my day job, I have written code that applies causal inferences to data stored in advertising logs. In my volunteer role on a DataKind project with the NYC Parks Department, I had to assess whether proactively pruning overgrown trees reduces the likelihood of a tree being hazardous during a storm. After some exploration of the problem, I realized this is the same problem as measuring an ad’s effectiveness. With about an hour’s worth of editing I adapted my Dstillery code to NYC Parks code (with my employer’s permission of course), and I effectively managed to answer the question at hand (YES, pruning does help). My success in helping the Parks Department was driven both by my willingness to commit beyond the initial weekend DataDive, and recognizing that I had solved that exact problem before (albeit with different data).
The automated data scientist doesn’t exist quite yet (though some are likely working on it), so enabling nonprofits to leverage data science the same way companies do will require a bit of labor at the outset. The initial investment in building the right tools and teaching data-driven methodology will continue to pay dividends long after the pro bono data scientist is gone. And the data scientist will gain so much as well – trust me, I know. The learning goes both ways on these types of engagements and there’s nothing quite like getting to flex your skills in an entirely new way. With the will and generosity of many good data scientists and the companies that employ them, such investments in the social sector are becoming more and more frequent. The results are inspiring as we’re seeing the applications of data science to improve our world are endless. See how your skills can help a nonprofit further its work addressing critical social issues – sign up to volunteer with DataKind.