Paper Title
Data Science Techniques for Analyzing Note Trends and Brand Progression in The Perfume Industry

Abstract
In the modern perfume industry, limited feedback on products can obstruct accurate market analysis and trend forecasting. This paper introduces an alternative rating strategy using a dataset from Fragnatica, which reveals significant gaps in available ratings for a wide range of perfumes. To address these inconsistencies, we developed a comprehensive pipeline using data science techniques for cleaning, filtering, and clustering fragrance data. By applying text analysis methods like Word2Vec and cosine similarity, we measured relationships between fragrance notes and employed K-Means clustering to group them into coherent categories. Cluster weights were calculated based on occurrence of fragrance notes, while designer performance was assessed by examining how often their perfumes appeared in the dataset.This approach offers deeper insights into fragrance trends and designer influence, providing a better understanding of market dynamics and product evaluations. Keywords - Data Science, Natural Language Processing, Perfume Industry, Trend Analysis, Text Similarity, Clustering.