make_spirals

make_spirals generates a synthetic data set composed of interlaced Archimedean spirals.

import matplotlib.pyplot as plt
X, y = make_spirals(random_state=0)
plt.scatter(X[:,0], X[:,1], c=y, cmap=plt.cm.bwr, alpha=0.5)

Install make_spirals using pip:

pip install make_spirals

make_spirals allows its customization. With n_samples (default 500) you control total number of points equally divided among classes and with noise (default 1) standard deviation of Gaussian noise can be added to the data.

import matplotlib.pyplot as plt
X, y = make_spirals(n_samples=1000, noise=2, random_state=0)
plt.scatter(X[:,0], X[:,1], c=y, cmap=plt.cm.bwr, alpha=0.5)

Using margin (default .5) you define then separation between each spiral.

import matplotlib.pyplot as plt
X, y = make_spirals(n_samples=1000, noise=2, margin=1.5, random_state=0)
plt.scatter(X[:,0], X[:,1], c=y, cmap=plt.cm.bwr, alpha=0.5)

By setting n_loops (default 2) you fix the number of loops of each spiral.

import matplotlib.pyplot as plt
X, y = make_spirals(n_samples=1000, n_loops=4, random_state=0)
plt.scatter(X[:,0], X[:,1], c=y, cmap=plt.cm.bwr, alpha=0.5)

Finally, n_classes (default 2) determines the total number of classes (i.e., spirals) to include in the dataset.

import matplotlib.pyplot as plt
X, y = make_spirals(n_samples=1000, n_classes=4, margin=1, n_loops=1, random_state=0)
plt.scatter(X[:,0], X[:,1], c=y, cmap=plt.cm.viridis, alpha=0.5)