Skip to content

Dataset · criteo

Criteo Uplift Modeling Benchmark

criteo/criteo-uplift

A large tabular dataset by Criteo AI Lab for evaluating and training uplift modeling models.

What's inside

Contains tabular data (10 million to 100 million rows) intended for binary classification and uplift modeling tasks. The dataset was published by Criteo AI Lab and LIG researchers at the AdKDD 2018 workshop as a large-scale benchmark.

What you can do with it

  • Training models for target action lift prediction (uplift modeling)
  • Comparing causal inference algorithms on large tabular data
Tasks
Tabular classification
Size
10M–100M rows
Downloads this month
3K
Likes
8

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("criteo/criteo-uplift")