Sales Prediction Using Machine Learning and Python

Abstract

In today\'s competitive market, being able to accurately predict sales is very important for managing inventory, planning revenue, and making strategic decisions. This project shows how to use machine learning to predict the sales and revenue of cricket products in real time. Brand, Product, Price, and Discount are some of the features used in this study. All of these have a direct effect on how customers buy things. We created and tested several machine learning models, such as Linear Regression, Decision Tree Regressor, Random Forest Regressor, and Support Vector Regressor (SVR). We chose SVR as the best model because it was better at predicting sales trends than the others.The project includes both console-based interaction and a graphical user interface (GUI) using Tkinter, which makes it easy for business owners and shopkeepers to use. The GUI lets users enter information about a product and see predicted sales and graphical analysis right away.The project was turned into a standalone .exe application so that the solution could be used on different computers without having to install Python. This project not only automates sales forecasting, but it also gives you information that can help you set better prices and discounts.

Introduction

Objective

The project aims to develop a machine learning-based system that predicts sales volume and revenue of cricket products based on inputs like brand, product type, price, and discount. It includes:

• A Python-based backend with ML models.

• A user-friendly GUI (built using Tkinter).

• A .exe version for offline use without Python.

Problem Statement

Retailers in the cricket equipment market struggle to predict sales accurately due to various influencing factors (e.g., price, brand). Without data-driven tools, they often overstock or understock. This project addresses that need with a smart predictive tool.

Project Goals

• Build a predictive model using features: Brand, Product, Price, Discount.

• Test ML models: Linear Regression, Decision Tree, Random Forest, SVR.

• Choose the best model (SVR).

• Design a console and GUI interface for input and results.

• Deploy the tool as a standalone desktop app.

Literature Review Highlights

• Traditional models (like ARIMA) require time-series data and don’t work well with categorical variables.

• ML models like Random Forest and SVR are better for structured, categorical datasets.

• SVR was chosen for its strong performance on small datasets with mixed data types.

• The project fills a gap by offering a lightweight, usable tool for small businesses using limited input features.

Background & Motivation

• Cricket is widely popular, making its gear a high-demand retail product.

• Shopkeepers often rely on guesswork for inventory and pricing.

• There’s a need for simple, offline, predictive tools that help small store owners make better business decisions.

System Design

Three-layer architecture:

1. User Interface: Console + GUI (Tkinter) with dropdowns, validation, and a real-time graph.

2. Application Logic: Processes inputs, loads the trained model, calculates and displays sales & revenue.

3. ML Model Layer: Models trained on encoded data; SVR selected using R² score and RMSE.

Implementation

• Developed in Python.

• Categorical data encoded.

• Multiple models tested and compared.

• Final model saved with joblib.

• GUI built with Tkinter, graph with matplotlib.

• Exported as .exe for offline access.

Key Benefits

• Accurate sales predictions using minimal input.

• Improved inventory and pricing decisions.

• Easy-to-use interfaces for non-technical users.

• No internet or coding needed to run the final tool.

Conclusion

Using machine learning, this project made a Cricket Product Sales Prediction System that works. The goal was to guess how many items would sell and how much money they would make based on things like the brand, type of product, price, and discount. Some of the most important things that happened are: Using more than one ML model, such as Linear Regression, Decision Tree, Random Forest, and SVR. We chose SVR as the best model because it was accurate and could work with data that wasn\'t linear. We made an easy-to-use interface with Tkinter GUI and Console-based interface. Real-time predictions with graph outputs made it easier for people who aren\'t tech-savvy to use (like shopkeepers). The project was put together into a single executable file (.exe) that can be easily installed on any Windows system. With this system, store owners and retailers can make better decisions about what to stock, plan sales, and predict sales revenue with confidence based on data.

Copyright

Copyright © 2025 Viyyapu Venkata Raviteja, Kuppili N Sathya Chitra. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.