DiffPriv 1.0.0b0

A Differential Privacy Package

The truth is more important than ever—let's make sure easy privacy protection is available.

Differential privacy should be simple. Now that data defines our world, we need to look at the cost of privacy. Let's make protecting privacy easy.

What is differential privacy?

Differential privacy allows for data to be preserved while making sure that attackers cannot gain access to an individual's data. Even if you publish summary statistics (like average age of participants, unlabeled addresses of participants, etc.), attackers can gain access to individual data (like age of each participant, labeled addresses of participants, etc.). In order to achieve this, differential privacy slightly changes the actual dataset to make sure that any uncovered data will not give away personal information. See below for how to get started!

Downloading DiffPriv

To download, open up your command prompt and type

    pip install DiffPriv==1.0.0b0 # This is a pre-release, so you need the version number

or from the source repo:

    git clone https://github.com/Quantalabs/DiffPriv
    cd diffpriv
    python setup.py install

Conda Envioronment

We currently do not have our package on Anaconda, however, we are working on getting it on conda, and should be available soon. However, there is a workaround for conda systems. Try building from the source with:

git clone https://github.com/Quantalabs/DiffPriv
cd DiffPriv

Then, create a conda virtual environment which should initialize pip with:

conda update conda
conda create -n DiffPriv python=3.9 anaconda
conda activate DiffPriv

Lastly, install dependencies with:

conda install numpy
pip install luddite  # luddite is not available on conda

Now, you can build the package from the source with:

conda setup.py install