The rapid growth of e-commerce has created many opportunities for online businesses, but it has also brought challenges such as product returns and unpredictable sales demand. Product returns can increase operational costs, affect customer satisfaction, and reduce business efficiency. At the same time, accurate sales quantity prediction is important for inventory planning, stock management, and revenue growth. This project analyzes an e-commerce transaction dataset to study return behavior and sales quantity prediction. The project focuses on two main machine learning tasks: classification and regression. The classification task aims to predict whether an order will be returned, while the regression task aims to predict the quantity of products sold. These analyses can help e-commerce companies identify return drivers, improve decision-making, reduce unnecessary costs, and manage stock more effectively.