roc-utils

This Python package provides tools to compute and visualize ROC curves. ROC curves can be used to graphically assess the diagnostic ability of binary classifiers.

Installation:

pip install roc-utils

To quickly test the installation, use the following cals

python -c "import roc_utils; print(roc_utils.__version__)"
python -c "import roc_utils; roc_utils.demo_bootstrap()"

Usage:

import numpy as np
import matplotlib.pyplot as plt
from roc_utils import *

def sample_data(n1, mu1, std1, n2, mu2, std2, seed=42):
    rng = np.random.RandomState(seed)
    #  sample size, mean, std
    x1 = rng.normal(mu1, std1, n1)
    x2 = rng.normal(mu2, std2, n2)
    y1 = np.zeros(n1, dtype=bool)
    y2 = np.ones(n2, dtype=bool)
    x = np.concatenate([x1,x2])
    y = np.concatenate([y1,y2])
    return x, y

x, y = sample_data(n1=300, mu1=0.0, std1=0.5,
                   n2=300, mu2=1.0, std2=0.7)
pos_label = True
roc = compute_roc(X=x, y=y, pos_label=pos_label)
plot_roc(roc, label="Sample data", color="red")
plt.show()

See examples/tutorial.ipynb for a more detailed introduction.

Build

To build the package, use the following

git clone https://github.com/hirsch-lab/roc-utils.git
cd roc-utils
python setup.py sdist bdist_wheel
python tests/test_all.py
python examples/examples.py