Sales Forecasting Project

Overview

The Sales Forecasting Project is designed to predict the future sales quantities of products using historical sales data. By leveraging machine learning techniques, specifically LSTM (Long Short-Term Memory) networks, this project aims to provide accurate forecasts to help businesses manage inventory and optimize their sales strategies.

Table of Contents

1. Project Structure
2. Installation
3. Usage
4. Data Description
5. Model Description
6. Results
7. Contributing
8. License

Project Structure

• Dataset.csv: The dataset containing historical sales data.
• Sales.py: Script for sales forecasting without additional parameters.
• Sales_Coupon.py: Script for sales forecasting considering coupons.
• Sales_Return.py: Script for sales forecasting considering returns.
• model.py: Script for sales forecasting considering both coupons and returns.
• README.md: This file.
• requirements.txt: List of Python packages required for the project.

Installation

1. Clone the repository:

   git clone https://github.com/your-username/Sales-Forecast.git
   
   

Usage

Each script is designed to forecast sales based on different parameters. You can run any of these scripts depending on the requirement:

To run the script without additional parameters:

python Sales.py

To run the script considering coupons:

python Sales_Coupon.py

To run the script considering returns:

python Sales_Return.py

To run the script considering both coupons and returns:

python model.py

Data Description

The dataset (Dataset.csv) contains the following columns:

• item_name: Name of the item.
• item_brand: Brand of the item.
• item_main_category: Main category of the item.
• item_sub_category: Sub-category of the item.
• transaction_date: Date of the transaction.
• item_quantity: Quantity of items sold.
• return_quantity: Quantity of items returned (used in Sales_Return.py and model.py).
• item_coupon: Coupon applied on the item (used in Sales_Coupon.py and model.py).

Model Description

The model used in this project is an LSTM (Long Short-Term Memory) network. The architecture includes:

• Two LSTM layers with ReLU activation.
• Dropout layer to prevent overfitting.
• Dense layer to output the prediction.

Results

The model's predictions are visualized and compared with the original sales data to assess its accuracy. The forecasted sales quantities help in understanding future sales trends.

Contributing

Feel free to open issues or submit pull requests if you have any improvements or suggestions.

License

This project is licensed under the MIT License.