This study presents a novel approach to predicting and analyzing performance problems in standalone Virtual Reality (VR) devices through the development of a comprehensive synthetic dataset and machine learning methodology. The research created a synthetic dataset simulating the performance of ten different standalone VR devices, incorporating both technical specifications and real-time performance metrics. The dataset generation process considered realistic device behavior patterns, including temperature variations under different load conditions, performance degradation factors, and network-related issues. The methodology employed seven different machine learning models. The dataset comprised 14,400 samples, with data collected at 5-second intervals over 120-minute sessions. Results demonstrated exceptional performance from tree-based models, with Random Forest and Decision Tree achieving near-perfect accuracy (99.97%). Extreme Gradient Boosting (99.69%) and Neural Network (98.92%) also showed strong performance. The study found that Overheating and Packet Loss predictions were particularly accurate across most models, while High Latency classification proved more challenging for some algorithms due to class imbalance. The synthetic dataset and methodology offer a foundation for future research in VR system optimization and real-time performance monitoring. The study addresses a significant gap in the literature by integrating both hardware specifications and performance metrics into a comprehensive analysis framework.