The project's goal was to transform raw, hard-to-use COVID-19 data from Johns Hopkins University into a clean, queryable dataset.
The initial daily CSV files were unusable for a stable API because their schemas, column names, and date formats changed over time.
The time-series CSV files provided a more stable and predictable data source with one column per day, which was better but still needed transformation.
A Python script was used to pivot the wide CSV data, transforming date columns into rows and creating a GeoJSON structure for mapping.
The final dataset is organized into five collections, including combined global and US data, pre-calculated daily counts, and a country summary collection.
MongoDB Atlas provides managed database clusters, while MongoDB Realm offers a backend-as-a-service layer for building serverless applications on top of Atlas.
MongoDB Realm can automatically generate a complete GraphQL API endpoint based on a collection's defined schema and access rules.
Realm provides built-in authentication providers, such as anonymous or email and password, to secure API endpoints and manage user access.
A custom REST API can be built using MongoDB Realm's HTTP services and webhooks, which execute serverless JavaScript functions to query the database.
Serverless platforms improve energy efficiency by sharing compute resources and eliminating idle servers, contributing to a greener data infrastructure.
MongoDB's flexible document schema allows for easier data migration and application updates by accommodating different document structures within the same collection.
28:16Alexis Yushin
28:12Gregor Bauer
57:44Sani Yusuf
46:47Ruwan Xaviour Fernando
44:51Stanimira Vlaeva
40:45Josh Goldberg
45:26Miki Lombardi
44:13Andrew Morgan
Jobs that call for the skills explored in this talk.
doinstruct Software GmbH
Berlin, Germany
CONTIAMO GMBH
Berlin, Germany
Peter Park System GmbH
München, Germany
MongoDB
Frankfurt am Main, Germany
MongoDB
Municipality of Madrid, Spain