SEPHORA PRODUCT ANALYTICS

Personal project | End-to-end data science | 2026

An end-to-end data science project built on the Sephora Products and Reviews dataset. Instead of one large notebook, the work is split into five independent notebooks, each answering a different business question on the same data: exploratory analysis, product success prediction, product clustering, price versus rating regression and a bonus "hidden gems" analysis. The analytical work is complemented by Power BI reports prepared for a business audience.

Key contributions:

Carried out exploratory data analysis across products, categories, prices and review volumes to establish what the catalogue actually looks like.

Built a classification model to predict whether a product will be highly rated, and an unsupervised clustering model to group products into meaningful segments.

Ran a regression analysis to test whether paying more leads to higher customer satisfaction.

Identified underrated, highly rated products with low review volume as merchandising opportunities.

Packaged the insights into three Power BI reports (EDA, product categories, hidden gems), delivered as PBIX files and PDFs for stakeholders.

Collaborators: Gabriela Słomiany

Software Used: Jupyter Notebook, pandas, scikit-learn, Power BI

Programming Languages: Python, DAX

Repository: github.com/imgabrielas/Sephora

Project Date: August 2026