RandomRBFDrift

Random Radial Basis Function generator with concept drift.

This class is an extension from the RandomRBF generator. Concept drift can be introduced in instances of this class.

The drift is created by adding a "speed" to certain centroids. As the samples are generated each of the moving centroids' centers is changed by an amount determined by its speed.

Parameters

• seed_model

  Type → int | None

  Default → None

  Model's random seed to generate centroids.

• seed_sample

  Type → int | None

  Default → None

  Sample's random seed.

• n_classes

  Type → int

  Default → 2

  The number of class labels to generate.

• n_features

  Type → int

  Default → 10

  The number of numerical features to generate.

• n_centroids

  Type → int

  Default → 50

  The number of centroids to generate.

• change_speed

  Type → float

  Default → 0.0

  The concept drift speed.

• n_drift_centroids

  Type → int

  Default → 50

  The number of centroids that will drift.

Attributes

• desc

  Return the description from the docstring.

Examples

from river.datasets import synth
dataset = synth.RandomRBFDrift(seed_model=42, seed_sample=42,
                               n_classes=4, n_features=4, n_centroids=20,
                               change_speed=0.87, n_drift_centroids=10)
for x, y in dataset.take(5):
    print(x, y)

{0: 1.0989, 1: 0.3840, 2: 0.7759, 3: 0.6592} 2
{0: 1.1496, 1: 1.9014, 2: 1.5393, 3: 0.3210} 0
{0: 0.7146, 1: -0.2414, 2: 0.8933, 3: 1.6633} 0
{0: 0.3797, 1: -0.1027, 2: 0.8717, 3: 1.1635} 0
{0: 0.1295, 1: 0.5953, 2: 0.1774, 3: 0.6673} 1

Methods

take

Iterate over the k samples.

Parameters

• k — 'int'