DT-MLP BASED FEATURE EXTRACTION MODEL FOR FRAUD AND LATE DELIVERY PREDICTION IN SCM
This paper presents a novel approach to detect and prevent fraud and late delivery in supply chain management using a hybrid model. Supply chain fraud and late delivery are significant issues that can lead to financial losses and damage to a company's reputation. Traditional methods for detecting fraud and late delivery, such as rule-based systems, can be time-consuming and may not be able to identify all instances of fraud and late delivery. Machine learning algorithms have been shown to be effective in detecting fraud and late delivery, but they can have limitations when applied to complex and dynamic supply chains. A hybrid model that combines multiple machine-learning techniques can improve the robustness and accuracy of predictions. In this study, a decision tree algorithm, and a multi-layer perceptron (MLP) algorithm were used to extract features from the data. The decision tree algorithm is useful for identifying patterns in the data, while the MLP algorithm is used to identify complex relationships between the features. Combine the two algorithm-extracted features for classification purposes. Our proposed novel hybrid solution shows that these extracted features combine (DT-MLP) with a logistic regression algorithm to classify fraud detection and late delivery prediction. The performance of the model was evaluated using various metrics such as accuracy, recall score, and F1 score. The machine learning algorithms were trained on large datasets containing historical data on fraud transactions, late delivery of orders, sales revenue, and quantity of products that customers order. The results show that the proposed hybrid model achieved exceptional performance with an accuracy of 99%. This approach can be a valuable tool for supply chain management to improve the efficiency, security, and transparency of the supply chain. With the help of this approach, organizations can identify potential fraud or late delivery early on, and take action to prevent it from happening, this will help them to save significant amounts of money in the long run.
Multi-layer perceptron (MLP), Decision tree (DT), Logistic regression, Random Forest, KNN