Predicting Sephora Product Popularity with Linear Regression and XGBoost

DOI

10.2991/978-94-6239-801-6_147How to use a DOI?

Keywords

Sephora; Product popularity prediction; Machine learning; XGBoost; Feature importance

Abstract

This project investigates whether product-level information can predict the popularity of Sephora products. The target variable is the loves count, defined as the number of users who have saved or expressed interest in a product. Because the raw loves count is heavily right-skewed, the analysis models log (loves count + 1). The data are split into 80% training and 20% validation. Linear regression and XGBoost are compared using mean squared error (MSE) and R2 on the log-transformed target. Linear regression achieves a validation MSE of 1.7998 and an R² of 0.4503. XGBoost achieves a lower MSE of 1.1018 and a higher R² of 0.6635. These results suggest that nonlinear relationships and feature interactions are important for predicting product popularity. The improvement from linear regression to XGBoost indicates that tree-based models capture complex patterns more effectively. Feature importance analysis further reveals that product category, reviews, and variation type are among the most influential predictors. Therefore, product-level attributes alone provide meaningful predictive signals for popularity on e-commerce platforms.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Yueran Ling
PY  - 2026
DA  - 2026/10/23
TI  - Predicting Sephora Product Popularity with Linear Regression and XGBoost
BT  - Proceedings of the 2026 2nd International Conference on Financial Innovation and Marketing Management (FIMM 2026)
PB  - Atlantis Press
SP  - 1409
EP  - 1414
SN  - 2352-5428
UR  - https://doi.org/10.2991/978-94-6239-801-6_147
DO  - 10.2991/978-94-6239-801-6_147
ID  - Ling2026
ER  -