Open source management#
How to install and run Cider#
For more information visit CIDER’s documentation below or github repo.
Roadmap for additional features#
Encryption/Decryption upon consent: We are adding functionality for personally identifying data to be encrypted prior to consent and only decrypted for analysis after a subscriber has consented. We are also building functionality for the encryption keys to be held by a third party for maximum security and auditing transparency.
Individual explanations of predictions: We plan to implement model-agnostic prediction explanations for mobile phone-based poverty estimates with LIME.
Ground rules for maintainers and contributors#
Review committee#
Bug reports, documentation, and code contribution will be reviewed by Emily Aiken (UC Berkeley) and Lucio Melito (GiveDirectly).
How a contribution is reviewed and accepted#
To report a bug, create an issue template on GitHub. Please include a full working example for the bug, including synthetic data if necessary.
To add additional documentation, create a new branch of Cider’s documentation repo, with
git checkout -b name-of-new-branch
. Commit your changes and push your new branch to the github repo, and create a pull request with the outline of your changes. Cider can always use additional documentation, so we appreciate your help!To write code, please create a new branch of Cider with
git checkout -b name-of-new-branch
. Commit your changes and push your new branch to the Github repo. Run the tests as indicated in the readme of the repo. Then, create a pull request. In your pull request, please include a description of all changes. For each change, include the files and line numbers that have been changed.
The types of contributions we’ll accept#
We will accept contributions to any of the modules in the code; we are particularly interested in engineering additional features and improving machine learning pipelines.
When it’s an appropriate time to follow up#
You can expect a response to an issue or suggestion within three months. We spend about one hour per month on this project.
Methodology used to build this open-source guide#
We modeled this guide on nine related open source guides:
Thinking Machines Data Science Poverty Mapping open source code
CommCare: supports frontline workers in low-resource settings [Link 2]
SORMAS: Surveillance, Outbreak Response Management and Analysis System [Link 2]
OCHA-Bucky: A COVID-19 model to inform humanitarian operations
Bandicoot: An open-source python package to analyze mobile phone metadata