Amazon Product Success Analysis

CAP4770 - Introduction to Data Science

Team Name: Group 4

Team Members

• Kaden Jones
• Louiza Joya
• Roshan Mehta
• Jacob Ramos

Synopsis

The goal of this project is to analyze and predict what makes a product successful on Amazon. This real world problem impacts sellers, businesses, and consumers on one of the largest e-commerce platforms. We will use techniques taught in class including classification and regression for prediction, and clustering for identifying patterns, and data visualizations to demonstrate findings.

Dataset

• Source: [Kaggle Amazon Products Dataset 2023] (https://www.kaggle.com/datasets/asaniczka/amazon-products-dataset-2023-1-4m-products?select=amazon_products.csv)
• Size: 1.4 million products
• Features: star ratings, reviews, price, productID, best seller

Project Structure

cap4770_finalproject_group4/
    model_comparison.py
    utils.py
cap4770_finalproject_group4/data/
    amazon_products.csv
    amazon_categories.csv
cap4770_finalproject_group4/notebooks/
    preprocessing.ipynb
    exploratory_data_analysis.ipynb
    xgboost_model.ipynb

Tools & Technologies

• Python
• Pandas
• Matplotlib
• Jupyter Notebook
• NumPy
• seaborn
• XGBoost
• scikit-learn

Installation

• Step 1: Clone the repo
• Step 2: Download the data and ensure it is within the /data folder. (see download_data.sh)
• Step 3: Install dependencies (see requirements.txt)
• Step 4: Run notebooks in order starting with preprocessing.ipynb

Evaluation Metrics

• Precision
• Recall
• Confusion matrix with sensitivity and specificity