Q: What qualifies as a Data for Good project that is executed with the volunteers or educational
groups?
A: CorrelAid has an excellent guide for understanding what makes a good Data for Good project.
https://docs.correlaid.org/project-manual/project-decision-guide. Alternatively, you may want to
check out the following blog post by the DSSG Summer Fellowship team: What Makes a Good Data
Science for Social Good Project? https://www.dssgfellowship.org/2015/11/04/what-makes-a-gooddssg-project
Q: How can I assess that an organization has sufficiently many data for running a Data for Good
project?
A: While sometimes Data for Good project may be a consultancy for organizing data collection, a
useful first step in understanding how mature is an organization with respect to data could be to
apply some of the data Maturity frameworks. One such Data Maturity framework is available from
Data Orchard: https://www.dataorchard.org.uk/resources/data-maturity-framework. Another such
framework is available from the DSSG Foundation (http://www.datasciencepublicpolicy.org/ourwork/tools-guides/datamaturity/). These frameworks can help you build an idea of an organization’s
technological and data readiness, but always check with a senior data scientist individual cases of an
organization.
Q: We are an NGO, an intergovernmental organization, or a public agency. Who can we partner
with for a Data for Good project?
A: There are many data scientists that are happy to invest their time to help good causes. If you have
a preference towards more organized teams, the options remain reduced. There are volunteer
organizations, either formal as associations or informal, that are accepting projects on an ongoing
basis, like CorrelAid and their X organizations, as well as DSSG Berling or DSSG Portugal. Other
options are university affiliated programs, like DSSG Summer Fellowships, that have specific timings,
but are also usually intense programs
Q: What does a typical Data for Good project team look like?
A: Team structure may vary depending on the type of the project. As Data for Good projects are
usually two-fold, on the one hand aiming to help a partner organization apply data science to
improve their operations and on the other hand to educate aspiring data scientists, as a minimum
there is a pair of a senior professional and a junior person with a set of skills that correspond to the
needs of a project.
A typical team for a project conducted within the DSSG summer fellowship is: 1 project manager, 1
senior technical mentor, and three or four junior data scientists. (Source:
A typical team for a volunteer-based project in DSSG chapters is comprised of a project lead and up
to 5 makers. (Source: https://www.dssg.pt/en/how-we-work/)
CorrelAid defines the following team roles: Project Coordinator, Team Lead, Team member, and they
reserve one spot in a project team for a “trainee” role. They usually have 2 to 8 data scientists in the
project, who are carefully selected from the list of applicants. (Source:
https://docs.correlaid.org/project-manual/project-coordinators/team-selection)
Other type of volunteer-based projects may feature hackathons where an objective is given and the
volunteers collaborate or compete to create a solution, where teams are self-organized and the
interaction with partner organizations is transactional and very short. But most Data for Good
projects require significant time dedication to reach a point where a positive effect is likely to occur.
Note that for each of the project teams approaches, Data for Good organizations emphasize
deliberation about ensuring diverse and inclusive teams.
Q: How to scope a Data for Good project?
A: Project scoping is a crucial task for any Data for Good initiative / organization. CorrelAid estimates
scoping to last up to 4 weeks, with several interactions with project partners, and offers tools for
their member to complete the task (https://docs.correlaid.org/project-manual/project-coordinators/scoping). Data Science for Social Good Foundation offers a Data Science Project
Scoping Guide with several templates to help this process:
http://www.datasciencepublicpolicy.org/our-work/tools-guides/data-science-project-scoping-guide/
(source: DSSG Foundation)
Q: What is a typical Data for Good project lifecycle?
A: A typical project lifecycle in a volunteer-driven organization is shown below.
(source: DSSG PT)
If you are an education institution that is interested in running DSSG-like summer fellowship,
consider the DSSG summer fellowship curriculum here: https://github.com/dssg/hitchhikersguide/tree/master/sources/curriculum. The summer plan for the fellowship is available here:
https://github.com/dssg/hitchhikers-guide/tree/master/sources/dssg-manual/summer-overview.
Note that the DSSG summer fellowship assumes that each project will be a challenging modelling
project. The reality for many other volunteer organizations may be different, and the generic project
lifecycle presented above (DSSG PT) may apply better.
Q: What are the best practices for running a project?
A: Different organizations nurture different sets of best practices, although there are some overlaps.
For best practices recommended by CorrelAid, check here: https://docs.correlaid.org/project-manual/project-team/best-practices. DSSG summer fellowship github is a treasure trove which can
also provide a collection of best practices, which you can check out here:
https://github.com/dssg/hitchhikers-guide/tree/master/sources/curriculum
Q: We have ethical and legal concerns regarding a Data for Good project. What is a good way to
deal with these concerns when interacting with Data for Good organizations or volunteers?
A: Ethical use of data science and legal considerations, especially when dealing with personal data or
data that may introduce biases in decision-making, are critical for both partner organizations and the
project teams. In fact, if there is no capacity in neither of the involved entities to manage such
concerns, and it is clear that there is a need for such management, the project should be paused.
Mature volunteer organizations, such as DSSG Portugal and CorrelAid have Ethics Committees who
advised teams on these matters. Additionally, CorrelAid has a person appointed to advise on GDPRrelated matters. You can download a presentation on this topic from the DSSG’s github:
https://github.com/dssg/hitchhikersguide/blob/master/sources/curriculum/3_modeling_and_machine_learning/machine-learning/ethics-ML.pdf Also, the same organization made a collection of materials on this topic
available here: https://github.com/dssg/hitchhikers-guide/blob/master/sources/curriculum/skills/ethics_bias_fairness.md
CorrelAid additionally provides a document for volunteers about the Data Security:
https://docs.correlaid.org/project-manual/data-security-and-privacy#declaration-on-data-security
In case of a doubt, always consider consulting with a Data Protection Officer or legal firms that can
provide some advice on this important topic.
Q: What is a typical output of a Data for Good project, and how is a project handed over?
A: A typical output of a short-term (up to 6 months) Data for Good project is a prototype of a
working model that requires further development and testing. These prototypes are usually
delivered via a shared github repository. However, the code is not sufficient, and there is often a
handover workshop where the partners are given a proper explanation about the use and
maintenance of the code, and the path for testing and deployment of the tested model.
Q: What is a typical Data for Good organizational structure?
A: The organizational structures vary across countries and the initiatives. We have identified two
organizations that have publicly shared their structures. One is DSSG Portugal:
https://www.dssg.pt/en/how-we-work/. Another one is CorrelAid:
https://docs.correlaid.org/wiki/hitchhikers-guide#our-organization
Q: Do I need to a legal entity to run Data for Good projects?
A: No. Many groups of volunteers act without a legal entity. However, to receive a donation and
sometimes to enter a legal agreement, you will need a legal entity. CorrelAid provides a short story
about their chapters and the choices they make: https://docs.correlaid.org/correlaidx-manual/finding-a-legal-form
