Cosmetics Dataset

Context

Imagine standing in front of a shelf filled with cosmetics, each one promising glowing skin, youthful appearance, or a flawless finish. The reality, though, is that the effectiveness and safety of these products depend largely on their chemical composition. For most consumers, deciphering these ingredient lists is nearly impossible, making it difficult to make informed choices. This dataset was created to address this very problem by providing a way to understand and compare cosmetic ingredients on a deeper level.

This dataset is designed to empower consumers and developers alike by making ingredient information more accessible and understandable. It contains detailed ingredient lists for 1,472 cosmetic products available on Sephora. The goal is to create a content-based recommendation system where the ‘content’ refers to the chemical components of these cosmetics. By analyzing these ingredients, users can gain insights into which products may suit their needs, avoid potential allergens, and even discover new products that match their preferences.

Content

The dataset offers a comprehensive collection of cosmetic ingredient data, meticulously gathered and processed for use in machine learning models. It includes the following:

• Ingredient Lists: Detailed chemical compositions of 1,472 cosmetics from Sephora, ranging from skincare to makeup products. Each entry includes the full list of ingredients as listed on the product’s packaging.
• Word Embeddings: To facilitate deeper analysis, the ingredient lists are processed using word embeddings, a technique that converts text data into numerical vectors that machine learning models can easily interpret.
• t-SNE Visualization: To visualize the similarity between different cosmetic products, we use t-SNE (t-Distributed Stochastic Neighbor Embedding), a popular machine learning method for dimensionality reduction. This allows for the creation of an interactive map where similar products are clustered together based on their ingredients.
• Interactive Visualizations with Bokeh: Using the Bokeh library, we create interactive visualizations that allow users to explore the ingredient similarities between products. This tool is invaluable for developers and researchers working on recommendation systems, as well as for consumers looking to make informed decisions about their cosmetic purchases.

Potential Use Cases

• Content-Based Recommendation Systems: Build personalized recommendation systems that suggest cosmetics based on the chemical composition of products users have already liked or used.
• Ingredient Analysis: Researchers and developers can use this dataset to analyze trends in cosmetic ingredients, identify common allergens, or study the effectiveness of certain components.
• Consumer Education: Create tools and applications that help consumers understand the ingredients in their cosmetics, enabling them to make more informed choices.
• Product Development: Cosmetic companies can utilize this dataset to analyze competitors’ products, explore new ingredient combinations, or develop products that cater to specific skin types and concerns.