This project aims to forecast product sales on Amazon using historical data that includes multiple influencing factors such as marketing spend, discount rate, and customer engagement (measured by page views). The goal is to predict Units_Sold for each time period, enabling more informed inventory management, marketing planning, and discount optimization.
I are provided with historical sales data from Amazon, including key features such as Date, Page_Views, Marketing_Spend, Discount_Rate, and Units_Sold. Each row represents daily sales performance for a product or category over time. The dataset captures user engagement (via page views), promotional efforts (via marketing spend), and pricing strategies (via discount rate), all of which influence customer purchasing behavior. Major shopping events like Prime Day, Black Friday, and festive sales are indirectly reflected through spikes in marketing spend and discounts, making it essential to model their impact accurately. One challenge is understanding how different variables interact and influence demand, especially during high-traffic sales periods.
This project aims to analyze these factors to build a predictive model that estimates future sales and supports data-driven e-commerce decisions.
-
e-commerce forecasting project.ipynb – Main Jupyter notebook containing all data analysis, modeling, and forecasting steps.
-
/data/ecommerce_amazon_dataset.csv – Folder to store the dataset (ecommerce_amazon_dataset.csv)
- Python (3.x)
- Google Colab or Jupyter Notebook
•numpy
•pandas
•matplotlib
•seaborn
•plotly
•scikit-learn