Pyrite Overview

Pyrite from Startup.ML is an anomaly detection tool optimized for high-dimensional, heterogeneous (both categorical and continuous) datasets. Its time complexity is near linear and space complexity is quadratic to the number of dimensions.

Pyrite is designed for categorical features (e.g., city can contain 'San Francisco', 'Boston', etc.) however it can also handle numerical features (e.g., income, age, etc) through discretization.

Getting started

from pyrite import Pyrite
Location = r'births1880.csv'
pyrite_data = Pyrite(pandas.read_csv(Location))
score_vec = pyrite_data.score_dataset(50, 100)


Pyrite uses the following dependencies:

  • numpy
  • pandas
  • matplotlib

Once you have the dependencies installed, download the distro:

tar xvfz pyrite-0.1.2.tar.gz

Go to the Pyrite folder and run the install command:

cd pyrite
sudo python install

Please review the license terms before installing and using Pyrite.


Please email your questions and comments to Startup.ML.