Construction and optimization of vending machine decision support system based on improved C4.5 decision tree

Area of Science:

• * Artificial Intelligence
• * Machine Learning
• * Operations Management

Background:

• * Increasing market competition necessitates refined operational management for manufacturers.
• * Machine learning (ML), a branch of artificial intelligence (AI), offers significant application prospects in various systems.
• * Vending machines present a practical case for applying AI to enhance operational efficiency.

Purpose of the Study:

• * To construct a product classification model for vending machines using the decision tree algorithm.
• * To build a sales forecasting model for vending machines utilizing neural networks (NN).
• * To develop a theoretical framework for a decision support system (DSS) for vending machines.

Main Methods:

• * Decision tree algorithms (C4.5 and improved C4.5) were employed for product classification.
• * Backpropagation neural networks (BPNN) were utilized for sales forecasting.
• * Reinforcement learning was incorporated to optimize system performance and adapt to market changes.

Main Results:

• * The C4.5 algorithm achieved accuracy rates between 68% and 87%.
• * The improved C4.5 algorithm demonstrated comparable accuracy (67%-87%) with significantly reduced running times across datasets.
• * BPNN sales forecasts showed high accuracy, with predicted data curves closely matching actual data.

Conclusions:

• * The developed DSS framework, integrating ML models, enhances vending machine operations.
• * The study demonstrates the effectiveness of AI in improving efficiency and customer satisfaction.
• * Reinforcement learning integration promises greater system intelligence and adaptability in dynamic market environments.