Parallel Coordinates Plotting for Python parcoords

This package implements a parallel-coordinate plot, commonly used in machine learning, i.e. for hyperparameter visualization.

Installation

The parcoords package can be installed with pip:

pip install parcoords

You can also view the package on PyPI.

Usage

The following code produces the sample plot at the beginning of this page.

import matplotlib.pyplot as plt
from parcoords import plot_parcoords

# let's define some data to plot
values = [
    [0.1, 10, "smote", 0.78],
    [0.15, 9, "none", 0.46],
    [0.18, 8, "rand_over", 0.67],
    [0.01, 16, "rand_over", 0.84],
    [0.01, 15, "smote", 0.86],
    [0.015, 12, "smote", 0.87],
    [0.03, 12, "smote", 0.81],
    [0.12, 13, "none", 0.51],
    [0.07, 11, "rand_over", 0.7],
    [0.2, 14, "none", 0.55],
]

# plot the values as parallel-coordinate plot
plot_parcoords(
    values,
    labels=["lr", "# epochs", "sampling method", "accuracy"],
    color_field="accuracy",
    scale=[("lr", "log")],
    title="Parallel-Coordinates of the Hyperparameters",
)

plt.show()

License

This repository is licensed under the MIT-License.