Leveraging Machine Learning In ERP To Predict Supply Chain Disruptions And Enhance Marketing Agility

Abstract

Within the concept of Industry 4.0, Enterprise Resource Planning (ERP) is transforming into an intelligent network that should enable predictive decisions in all areas of the company. This research presents an integrated ML framework in the ERP system to predict SC disruptions and improve marketing agility. The architecture uses supervised and unsupervised learning approaches (e.g., Random Forest, Support Vector Machines (SVM), K-means clustering, and attention-based Long Short-Term Memory (LSTM) networks) to read multi-source data (supplier metrics, logistics data, social media sentiment, customer behavior). A data pipeline in real time transfers the internal ERP modules to external data streams to extract dynamic features, and to detect anomalies. Time-series and deep learning models are used for disruption prediction and they capture linear patterns as well as temporal dependencies. Meanwhile, via reinforcement learning, we optimize the marketing policy over the changing supply-demand state. Experimental results on synthetic ERP datasets show an accuracy of 92.7% in predicting disruption and 42% improvement in marketing response time over baseline ERP analytics. Explainer AI (XAI) modules are incorporated, to make model transparent and decisions traceable. The findings indicate that ML-enabled ERPs can produce substantive benefits on both operational robustness and dynamic adaptiveness within a complex context of turbulent environment.