prteek/ratings-compare

Headphone star ratings

7,681 star ratings of 46 headphones from three brands (Philips, Jabra and Apple) by 7,605 users, 2005–2018. This is the data behind the blog post Is One Product Really Rated Better? on Hugging Face, which compares how users rate two products with a beta-binomial model and profile likelihood, and the default example in the ratings-compare Space. Code: github.com/prteek/ratings-brand-analysis.

Source

A subset of the Electronics dataset released with Wan, Ni, Misra & McAuley (2020), Addressing Marketing Bias in Product Recommendations, WSDM (github.com/MengtingWan/marketBias, Apache-2.0). Kept: every rating in the Headphones category from the brands Philips, Jabra and Apple; all columns as released. This subset is shared under the same licence.

Columns

column	content
item_id	product id (46 products: Philips 30, Jabra 14, Apple 2)
user_id	anonymised user id, as in the source
rating	stars, 1–5
timestamp	date of the rating
model_attr	product attribute from the source: Male, Female or Female&Male
category	Headphones for every row
brand	Philips, Jabra or Apple
year	year from the source data
user_attr	user attribute from the source: Male or Female, missing for most rows
split	the source's train / validation / test split (0, 1, 2)

See the source paper for how model_attr, user_attr and split were derived.

The blog post compares Apple item 4501 (76 ratings) with Philips item 3700 (38 ratings). There, a rating of s stars is modelled as s − 1 favourable outcomes out of 4.

Citation

@inproceedings{wan2020marketing,
  author    = {Mengting Wan and Jianmo Ni and Rishabh Misra and Julian McAuley},
  title     = {Addressing Marketing Bias in Product Recommendations},
  booktitle = {Proceedings of the 13th International Conference on Web Search and Data Mining (WSDM)},
  year      = {2020}
}