A new paper in SoftwareX presents trasgoDP, the open-source library developed within EOSC-SIESTA for applying differential privacy to sensitive data. It is a modular Python framework for releasing tabular microdata, designed for integration within standard data science workflows.
trasgoDP implements classical differential privacy mechanisms adapted to the local setting. These mechanisms are applied at the record level, producing locally perturbed versions of the data with record-level privacy guarantees while preserving the data’s original structure. Unlike existing software focused on applying differential privacy to aggregated data, trasgoDP supports row-wise microdata and location data release. Among the libraries reviewed in the paper, it is the only one operating at the record level.
The main challenge is the balance between privacy and utility. Local differential privacy requires considerably more noise than the central model, which can substantially reduce the data’s utility; conversely, if the privacy budget (ε) is set too high, the guarantees become very weak. To guide the choice of privacy budget, the library enables the exploration of privacy-utility trade-offs across multiple mechanisms, data types and privacy budgets, with metrics (divergence measures and a novel correlation-loss metric) that quantify the information loss as the privacy level changes.
A key feature is a record-level metric-privacy mechanism, based on geo-indistinguishability, that works directly on longitude-latitude coordinates.
The aim is to offer a practical, easy-to-use tool that helps data practitioners realistically assess the feasibility and limits of local differential privacy in controlled environments, always taking into account the trade-off between privacy and utility. trasgoDP also enables reproducible experimentation and the generation of noisy datasets, and complements pyCANON and anjana as the third component of a toolkit for privacy-preserving microdata release.
trasgoDP is open source (Apache-2.0) and can be installed via pip install trasgodp. It can also be tested interactively through the SIESTA dashboard and at https://trasgodp.cloud.eosc-siesta.eu/.
Read the full paper (open access): J. Sáinz-Pardo Díaz and Á. López García, “trasgoDP: An open source framework for releasing noised tabular microdata under local differential privacy”, SoftwareX 36 (2026) 103171. https://doi.org/10.1016/j.softx.2026.103171