Miguel José Monteiro – Lead Team DSSG PT

By now, there is no question about the importance, power and value of data. Companies have already figured out how to make it profitable, be it through selling new products and services or simply through optimizing internal processes and saving money. However, there is not a lot of talk about the social impact of data and how it can directly help those in need and society as a whole.

Call it the new oil or the new gold, data is everywhere and is generated all the time. When asked about it, we can all come up with commercial examples of data products: the recommendation systems, the search engines, the credit risk algorithms. And these are all useful, no question about it — they solve a problem and they improve people’s lives in some way.

On the other hand, not a lot of thought (much less work) has been put on using these same algorithms to create non-commercial social impact by helping more underprivileged groups. The truth is these algorithms that companies use to boost profits can also be used by social good organizations — after all, being data-driven is not limited to companies. In fact, data science can also help governments, nonprofits and social good organizations run their operations more efficiently.

So why is there little work done? Well, it’s the usual reason — it is not profitable.

Most organizations that work in any kind of social change field don’t have the budget or staff to take full advantage of this data revolution. More than that, they don’t even understand how data can help them — and I don’t blame them, they already have their plates more than full with their day-to-day operations (you know, saving the world and stuff). And at the end of the day, these organizations survive mostly out of volunteer work that is not scalable in any way, shape or form. But what if we could show these organizations just how valuable data can be for them too? What if we could introduce scalability in their work using data?

Besides this, it is also not clear for data science professionals how they can help.

Most data scientists don’t realize just how valuable their skills can be and how impactful their work can become if their expertise was channeled in a different way than what they’re already doing on their job. But truth be told, there are a lot of areas where data science can have a direct social impact. From poverty and hunger alleviation, child well-being, healthcare and education access, to environmental causes, data science has the potential to truly move the needle on seemingly impregnable issues. However, this is only possible if there is a strong cooperation between data scientists and social sector experts.

So, what to do now?

Disappointed with the fact that the number of organizations tackling this issue is less than a handful worldwide, yet inspired by this societal need, me and a group of amazing people who share the same vision about data science and social impact got together and created Data Science for Social Good Portugal. We want to help social good organizations through the whole data processing pipeline — from data collection to predictive modelling — according to what they truly need.

To make things easier and faster, we partnered with DSSG Chicago (and borrowed the name along the way). They already have a bunch of social good projects on their portfolio — such as this one, about using data science to improve police interactions in several Police Departments in the United States. DataKind is another organization that’s been doing this for quite some time now (and who were nice enough to meet with us, share some of their practical knowledge and give hints on how to kick-start this project — thanks for that, guys!). One project they did tackled  homelessness in the United Kingdom and how data about vulnerable populations can decrease the prevalence of this problem.

Hopefully these two examples will give you a grasp on the potential of using data science for social good and will motivate you to look further on what you can do.

The English edition was initially published in the Towards Data Science Medium Publication