This project focuses on forecasting future sales using historical business data from a Superstore dataset. The goal is to analyze sales trends, build a forecasting model, and provide business-friendly insights that can help organizations make better inventory, staffing, and financial planning decisions.
- Predict future sales based on historical sales data.
- Perform data cleaning and preprocessing.
- Create time-based features for forecasting.
- Build and evaluate a machine learning model.
- Visualize sales trends and future forecasts.
Superstore Sales Dataset Dataset Source: https://www.kaggle.com/datasets/vivek468/superstore-dataset-final
The dataset contains 9,994 sales records and 21 features related to orders, customers, products, sales, and profits.
- Python
- Pandas
- NumPy
- Matplotlib
- Scikit-learn
- Jupyter Notebook
- VS Code
- Loaded and explored the dataset
- Checked and confirmed no missing values
- Converted Order Date to datetime format
- Created Year, Month, and Day features
- Aggregated monthly sales data
Linear Regression
- Year
- Month
- Sales
- Mean Absolute Error (MAE): 8524.89
- Root Mean Squared Error (RMSE): 10276.69
- R² Score: 0.52
- Monthly Sales Trend
- Actual vs Predicted Sales
- Future Sales Forecast
- Sales show an overall increasing trend over time.
- Peak sales occur during specific periods, indicating seasonal behavior.
- The forecast predicts continued growth in sales throughout 2018.
- Businesses can use these insights for inventory management, staffing decisions, budgeting, and demand planning.
This project demonstrates how machine learning can be used for sales forecasting and demand prediction. Historical sales data was analyzed, visualized, and used to build a forecasting model that provides actionable business insights.
Pravallika B.Tech CSE Student