Datasets:

relbench
/

core

task	kind	type	description
`item-churn`	forecast	binary_classification	Churn for a product is 1 if the product recieves at least one review in the time window, else 0.
`item-ltv`	forecast	regression	LTV (life-time value) for a product is the numer of times the product is purchased in the time window multiplied by price.
`review-rating`	autocomplete	regression	Predict the `rating` column of the `review` table.
`user-churn`	forecast	binary_classification	Churn for a customer is 1 if the customer does not review any product in the time window, else 0.
`user-item-purchase`	forecast	link_prediction	Predict the list of distinct items each customer will purchase in the next two years.
`user-item-rate`	forecast	link_prediction	Predict the list of distinct items each customer will purchase and give a 5 star review in the next two years.
`user-item-review`	forecast	link_prediction	Predict the list of distinct items each customer will purchase and give a detailed review in the next two years.
`user-ltv`	forecast	regression	LTV (life-time value) for a customer is the sum of prices of products that the customer reviews in the time window.

Loading

import relbench
ds = relbench.load_dataset("relbench/v1/rel-amazon")
task = relbench.load_task("relbench/v1/rel-amazon", "<task>")

Citation

Please cite RelBench:

@inproceedings{robinson2024relbench,
  title     = {{RelBench}: A Benchmark for Deep Learning on Relational Databases},
  author    = {Robinson, Joshua and Ranjan, Rishabh and Hu, Weihua and Huang, Kexin and Han, Jiaqi and Dobles, Alejandro and Fey, Matthias and Lenssen, Jan E. and Yuan, Yiwen and Zhang, Zecheng and He, Xinwei and Leskovec, Jure},
  booktitle = {Advances in Neural Information Processing Systems 37 (NeurIPS 2024) Datasets and Benchmarks Track},
  year      = {2024}
}