NewsArticles

News articles bandit dataset.

This is a personalization dataset. It contains 10000 observations. There are 10 arms, and the reward is binary. There are 100 features, which turns this into a contextual bandit problem.

Attributes

• arms

  The list of arms that can be pulled.

• desc

  Return the description from the docstring.

• is_downloaded

  Indicate whether or the data has been correctly downloaded.

• path

Examples

from river import bandit

dataset = bandit.datasets.NewsArticles()
context, arm, reward = next(iter(dataset))

len(context)

100

arm, reward

(2, False)

Methods

download

take

Iterate over the k samples.

Parameters

• k — 'int'

---

1. Machine Learning for Personalization homework ↩

2. Contextual Bandits Analysis of LinUCB Disjoint Algorithm with Dataset ↩