Data scarcity is a significant problem in Enterprise Resource Planning (ERP) adoption prediction, limiting the accuracy and reliability of traditional predictive models. This study addresses this issue by integrating Generative Artificial Intelligence (AI) technologies, specifically Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), to generate synthetic data that supplements sparse real-world data. A systematic literature review identified critical gaps in existing ERP adoption models, underscoring the need for innovative approaches. The generated synthetic data, validated through comprehensive statistical analyses including mean, variance, skewness, kurtosis, and the Kolmogorov-Smirnov test, demonstrated high accuracy and reliability, aligning closely with real-world data. A hybrid predictive model was developed, combining Generative AI with Pearson Correlation Coefficient (PCC) and Random Forest techniques. This model was rigorously tested and compared against traditional models such as SVM, Neural Networks, Linear Regression, and Decision Trees. The hybrid model achieved superior performance, with an accuracy of 90%, precision of 88%, recall of 89%, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC) score of 0.91, significantly outperforming traditional models in predicting ERP adoption outcomes. The research also established continuous monitoring and adaptation mechanisms to ensure the model's long-term effectiveness. The findings provide practical insights for organizations, offering a robust tool for forecasting ERP adoption success and facilitating more informed decision-making and resource allocation. This study not only advances theoretical understanding by addressing data scarcity through synthetic data generation but also provides a practical framework for enhancing ERP adoption strategies.
