# When Fashion Meets Big Data: Discriminative Mining of Best Selling Clothing Features

**Authors:** Kuan-Ting Chen, Jiebo Luo  
**arXiv:** [1611.03915](https://arxiv.org/abs/1611.03915)  
**Date:** 2016-11-11  
**Summary page:** https://www.summarizepaper.com/en/paper/1611.03915v2/

## Key points

- E-commerce dominance and online shopping convenience have led to a vast array of product options for consumers, particularly in clothing.
- Clothing is one of the most popular sectors in e-commerce, prompting analysis of key features that make certain items appealing to consumers.
- Kuan-Ting Chen and Jiebo Luo developed a novel system with three main components:
- Exploration and organization of a large-scale dataset from an online clothing platform
- Pruning and extraction of images of best-selling products along with user transaction history to identify popularity patterns
- Employment of a machine learning-based approach to discern fine-grained attributes specific to popular clothing styles
- The system's effectiveness is demonstrated through extensive experimentation on an online clothing shopping dataset, providing valuable insights into prevailing trends in clothing consumption and profitable features driving consumer interest.
- This research contributes to understanding consumer behavior within the context of online apparel shopping by combining fashion with big data.

## Summary

In the age of e-commerce dominance and the convenience of online shopping, consumers are faced with a vast array of product options, particularly in the realm of clothing. Shopping has become a fundamental activity in society, with significant implications for both industry practices and academic disciplines such as sociology and psychology. Among the various categories within e-commerce, clothing stands out as one of the most popular sectors, prompting an analysis of key features that make certain clothing items appealing to consumers. This not only aids in enhancing existing applications like clothing recommendations and advertising but also sheds light on emerging trends in consumer preferences. To address this demand, Kuan-Ting Chen and Jiebo Luo have developed a novel system comprising three main components. Firstly, it involves exploring and organizing a large-scale dataset sourced from an online shopping platform specializing in clothing items. Secondly, the system prunes and extracts images of best-selling products from this dataset along with user transaction history to identify patterns of popularity. Lastly, a machine learning-based approach is employed to discern fine-grained attributes specific to popular clothing styles, serving as representative markers for discriminative characteristics. Through extensive experimentation conducted on a substantial online clothing shopping dataset, this system demonstrates its effectiveness. Valuable insights are gleaned regarding prevailing trends in clothing consumption and identifying profitable features that drive consumer interest. By delving into the intersection where fashion meets big data, this research contributes to a deeper understanding of consumer behavior within the context of online shopping for apparel.