DriftRetrainingClassifier

Drift retraining classifier.

This classifier is a wrapper for any classifier. It monitors the incoming data for concept drifts and warnings in the model's accurary. In case a warning is detected, a background model starts to train. If a drift is detected, the model will be replaced by the background model, and the background model will be reset.

Parameters

• model

  Type → base.Classifier

  The classifier and background classifier class.

• drift_detector

  Type → base.DriftAndWarningDetector | base.BinaryDriftAndWarningDetector | None

  Default → None

  Algorithm to track warnings and concept drifts. Attention! If the parameter train_in_background is True, the drift_detector must have a warning tracker.

• train_in_background

  Type → bool

  Default → True

  Parameter to determine if a background model will be used.

Examples

from river import datasets
from river import evaluate
from river import drift
from river import metrics
from river import tree

dataset = datasets.Elec2().take(3000)

model = drift.DriftRetrainingClassifier(
    model=tree.HoeffdingTreeClassifier(),
    drift_detector=drift.binary.DDM()
)

metric = metrics.Accuracy()

evaluate.progressive_val_score(dataset, model, metric)

Accuracy: 86.46%

Methods

learn_one

Update the model with a set of features x and a label y.

Parameters

• x
• y
• kwargs

Returns

self

predict_one

Predict the label of a set of features x.

Parameters

• x — 'dict'
• kwargs

Returns

base.typing.ClfTarget | None: The predicted label.

predict_proba_one

Predict the probability of each label for a dictionary of features x.

Parameters

• x
• kwargs

Returns

A dictionary that associates a probability which each label.