Data For Good x Oxfam

Data For Good is a French community of data scientist & developers that put their skills to the benefit of solving social problems. First as a meetup, it is now a small acceleration program, matching & accompanying voluntaries with associations and NGOs twice a year.

I participated in the batch #2 (Autumn 2016) and chose to help Oxfam in analysing the role that tax havens play in the activities of the top 20 European Banks.

Since 2013, all European banks are required to publish information on their activities (profits & turnover), staff, tax paid and subsidies received in each territory in which they are established - including tax havens. In March 2016, a first report published by Oxfam (you can find it here) examined the country-by-country reportings of the five largest French banks (BNP Paribas, BPCE, Société Générale, Crédit Agricole, and Crédit Mutuel).

The goal is to publish a similar report in March 2017, but this time going much further in the analysis: cleaner data, enrich & cross with open data, more countries, extract & investigate their subsidiaries. We will also build a small website to let people visualize, understand & filter the data.

What we did

• extracted data from pdf tables with the open-source tool Tabula
• cleaned & processed data using Python & Pandas
• used Dataiku & PostgreSQL to centralize everything

Go further

The final report can be viewed here.