A Multiple Classifier System for Predicting Best-Selling Amazon Products

<p>In this work, I examine a dataset of Amazon product metadata and propose a heterogeneous multiple classifier system for the task of identifying best-selling products in multiple categories. This system of classifiers consumes the product description and the featured product image as input and feeds them through binary classifiers of the following types: Convolutional Neural Network, Na¨ıve Bayes, Random Forest, Ridge Regression, and Support Vector Machine. While each individual model is largely successful in identifying best-selling products from non best-selling products and from worst-selling products, the multiple classifier system is shown to be stronger than any individual model in the majority of cases of identifying best-selling products from non best-selling products, and achieves up to 83.3% accuracy, depending on the product category. To my best knowledge, this research is the first application of ensemble learning to Amazon product data of this type and the first use of product images and Convolutional Neural Networks to predict product success.</p>

Name thesis:degree_name

Master of Science in Computer Science (MSCS)

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Year dc:date.available

2018

Author dc:creator

• Kranzlein, Michael

Contributors dc:contributor

• Dan Chia-Tien Lo
• Mingon Kang

• data analysis
• ecommerce
• image classification
• machine learning
• text classification
• Other Computer Sciences

Repository record dc:identifier

https://digitalcommons.kennesaw.edu/cs_etd/18

OAI identifier oai:identifier

oai:digitalcommons.kennesaw.edu:cs_etd-1012