{"repo":"imperial-aisp/pycorrectmatch","free":true,"listed":false,"github":"https://github.com/imperial-aisp/pycorrectmatch","clone":"git clone https://github.com/imperial-aisp/pycorrectmatch.git","description":"Source code [Python library] for https://nature.com/articles/s41467-019-10933-3","language":"Python","stars":30,"topics":["privacy","privacy-protection","python","python-wrapper","statistics"],"license":"GPL-3.0","category":"api-integrations-sdks","readme_excerpt":"CorrectMatch A thin Python wrapper around the Julia module CorrectMatch.jl, to estimate uniqueness from small population samples. Installation Install CorrectMatch using your favorite package manager, e.g., pip: or uv: We use JuliaCall to seamlessly run Julia code from Python. The Julia binary and its dependencies, including CorrectMatch.jl, will be automatically installed on first use. Usage This module estimates the uniqueness of a population, on which multiple discrete attributes can be collected. For instance, the following array is a sample of 1000 rows for five discrete attributes: We can estimate the uniqueness of a population of 1000 individuals, or 10000 individuals, from this sample: by fitting a copula model to the observed records: CorrectMatch can also estimate uniqueness and correctness directly from pandas DataFrames, including those with categorical or string columns: Individual-level metrics Beyond population-level metrics, CorrectMatch can estimate the uniqueness and correctness of a specific individual given a fitted model in, say, a population of 1000 records: These functions estimate how likely a specific record is to be unique or correctly re-identified in a population. In the demo/ folder, we have compiled more examples with real-world datasets. License GNU General Public License v3.0 See LICENSE to see the full text. Patent-pending code. Additional support and details are available for commercial uses.","default_branch":null,"files":null,"tree":[],"storefront":"/r/imperial-aisp","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/imperial-aisp/pycorrectmatch/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}