Skip to content

PerOutputClassifier

A multi-output model that trains one independent classifier per output.

This model does not use the prediction of one output as a feature for the next. Each output is modelled by its own copy of the base classifier, trained independently. (This is the streaming equivalent of scikit-learn's MultiOutputClassifier).

The set of outputs isn't known from the start in a streaming setting, new classifiers are instantiated on the fly, one per output key encountered in y.

Parameters

  • classifier

    Typebase.Classifier

    A classifier model used for each label.

Examples

import random

from river import datasets
from river import feature_selection
from river import linear_model
from river import metrics
from river import multioutput
from river import preprocessing

dataset = list(datasets.Yeast())
random.Random(42).shuffle(dataset)

model = feature_selection.VarianceThreshold(threshold=0.01)
model |= preprocessing.StandardScaler()
model |= multioutput.PerOutputClassifier(
    classifier=linear_model.LogisticRegression(),
)

metric = metrics.multioutput.MicroAverage(metrics.Jaccard())

for x, y in dataset:
    y_pred = model.predict_one(x)
    y_pred = {k: y_pred.get(k, False) for k in y}
    metric.update(y, y_pred)
    model.learn_one(x, y)

metric
MicroAverage(Jaccard): 41.82%

Methods

learn_one

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

Parameters

  • x
  • y
  • kwargs

predict_one

Predict the labels of a set of features x.

Parameters

  • x
  • kwargs

Returns

The predicted labels.

predict_proba_one

Predict the probability of each label appearing given dictionary of features x.

Parameters

  • x
  • kwargs

Returns

A dictionary that associates a probability which each label.