This project generates utilizes differential privacy methods on the NYC taxi dataset to mask columns with sensitive data and deploy a private dataset. We utilize 5 methodologies: Total Anonymity, Min/Max Uniform Distribution Sampling, Gaussian Sampling, Laplacian Sampling, and GUPT. Please see the Google Drive link below for a project description and the GitHub link for our implementation. The Google Docs link is for the slideshow (only accessible from a Columbia email).
Incremental Generation of Differentially-Private Datasets
We explore 5 ways of generating data on yellow cabs that differentially-privatizes the original dataset. Methods build upon previous methods until we result in a stable dataset-generating algorithm.
Updates
Leave feedback in the comments!
Log in or sign up for Devpost to join the conversation.