This is a repository for investigating differential privacy mechanisms in transportation area. We implemented two privacy mechanisms for the sporadic case and an algorithm for the repeated case as described in the paper Geo-indistinguishability: A Principled Approach to Location Privacy. We demonstrated the mechanisms using NYC taxi trip data for sporadic case and Porto taxi trajectory data for repeated case.

For obfuscating single locations, we use the geo-indistinguishable mechanism of optimal utility proposed by the paper above with the aid of greedy graph spanner algorithm as a way to reduce the complexity. The code can be found in Greedy_Graph_Spanner.ipynb.

For obfuscating a whole trajectory, we implemented the predictive dX-private mechanism proposed by the paper. It's worth noting that in practice the prediction model can be chosen arbitrarily. In our demonstration, we built a simple trajetcory prediction model by combing two stacked LSTM layers together as detailed pkdd-15-predict-taxi-service-trajectory-i/Trajectory_Prediction.ipynb. The full code can be found in Trajectory_Privacy.ipynb. (please download the data from Kaggle if you are interested in running the code).