Skip to content

Dataset · zozonext

Bandit A/B test logs for clothing recommendations

zozonext/open-bandit

26 million user interaction records with ZOZOTOWN online clothing store recommendations for training bandit algorithms.

What's inside

The full version of the Open Bandit Dataset in tabular and text formats. Inside are 26 million rows of A/B test data from two multi-armed bandit algorithms collected on the major Japanese e-commerce clothing platform ZOZOTOWN. Designed for research and off-policy evaluation.

What you can do with it

  • Train and evaluate multi-armed bandit algorithms
  • Test recommendation systems via off-policy evaluation
  • Benchmark models for e-commerce recommendations
Size
10M–100M rows
Downloads this month
22.7K
Likes
1

How to load it

from datasets import load_dataset

ds = load_dataset("zozonext/open-bandit")