I have compiled a dataset of evangelical churches in Brazil. The dataset is publicly available (https://github.com/araujosvictor/igrejas_UF_100milhab
) and was extracted from Brazil’s federal tax authority (Receita Federal), using the initial registration information (CNPJ) required for any legally registered organization in Brazil. The dataset includes the full address provided at the time each church was opened, and I have already geocoded the records to add latitude/longitude coordinates. In total, the dataset contains 4,763 churches.
However, the dataset still requires cleaning. This includes relatively straightforward tasks (for example, removing entries that are not evangelical churches), as well as a more important step: validating the addresses. For each row, I plan to add a dedicated address-validation column coded as 1 if the address is correct—or if the church can be located in the immediate vicinity despite minor address inaccuracies—and 0 otherwise. This validation will be based on Google Maps searches and publicly available sources such as Facebook. I also plan to include an additional column to flag cases where a church with the same name exists, but at a different address. Accuracy is a priority throughout this process.