DataKind, Alteryx for Good, Tableau Foundation and VolunteerMatch.org are a few of the platforms where data scientists can connect with charities that want to benefit from their skills. Emma Deraze, a data scientist at TES Global, is a volunteer with DataKind and has worked on several charity projects, while consulting analyst with The Information Lab, Ben Moss, has used Tableau Foundation to connect with a US-based charity on a specific project.
At the Data and Tech for Good conference organised by Analysis Marketing, Deraze and Moss explained how those organisations worked and the effect they each had on a specific project. “Charities are trying to solve a problem and data is a very powerful asset to have, but the skills required to do analysis are really expensive skills,” sad Deraze.
To alleviate that financial burden, DataKind organises hackathons focused on charitable data analysis in six cities around the world. Two-day hackathons are known as DataDives, while those lasting a day are called DataJams. A few DataKind volunteers will prepare the data alongside the charity in the three months leading up to a hackathon, with 20 to 50 data specialists offering their knowledge and expertise during the session.
“The volunteer data scientists complete analyses that the charities could not have done in-house,” Deraze said. She gave an example of a DataDive that took place in October 2016 for the social organisation WeFarm with 80 volunteer data scientists. WeFarm is a peer-to-peer, information-sharing service that allows small-holder farmers in developing countries to communicate with each other, as well as ask questions and receive crowdsourced answers via SMS.
The challenge WeFarm faced was the communication network was not being used efficiently. Some people were using WeFarm as a social network, which frustrated other users causing them to stop using it. Also, questions were going unanswered because they were being directed to people who didn’t know about that specific topic or who didn’t speak the language the question was asked in.
DataKind identified the solution of having super-users who would be more willing to answer questions and be able to redirect questions that he or she couldn’t answer to users who could. “When you think of network efficiency generally, there are always nodes that are really important and, if you remove them, everything crumbles. We were trying to identify who those people were,” said Deraze.
To do this, the DataKind volunteers ran sentiment analysis, as well as natural language testing, and designed a prototype that pushed messages to active super-users. As a result of the hackathon, WeFarm had a better-organised system and was able to reach more people, which is essentially what Moss and some of his colleagues did for the charity Connect2Help211.
The connection was made through the Tableau Foundation, which Moss described as the Tinder for analytics people and charitable organisations. Connect2Help211 maintains a database of community and social services providers in the state of Indiana. Callers in need of basics like food or shelter ring the hotline and are referred to the service that can best support them.
The project began with a brief from the director of analytics at the charity, who primarily wanted to have the extraction, transformation and loading (ETL) process cleaned up. The first challenge Moss faced was difficulty in getting hold of the data, as the database was owned and run by a third party. “If Connect2Help wanted any data, they would contact the database admin within the external company who would manually query the data and send it back. It was a slow process,” said Moss.
The second challenge was the data was messy. They had used free-text fields and staff had found numerous ways to categorise the same thing. Moss fixed this by using Alteryx to connect into the referrer database and using Alteryx again to manipulate the false values and turn them into the actual values with a series of look-up tables. He would also reject any new, incorrect values that came in.
“The final part was outputting a series of tables to their MySQL database which they link and get live into Tableau to produce a series of data visualisations,” said Moss. He also advised them to stop using free text.
Moss said that charities face the same data analysis challenges that any other business would. He added that he and others like him are doing what they can to help meet those challenges. “There is a lack of data availability, the data is messy and they don’t know the best way to communicate that information. That’s why we’re so important as experts in the field - we can go into these organisations and help them,” said Moss.