Artificial Intelligence-Based Predictive Analytics Framework for Real-Time Decision Support in Smart Computing Environments

Abstract

The rapid evolution of smart computing environments has generated massive volumes of heterogeneous data from Internet of Things (IoT) devices, cloud platforms, edge computing systems, cyber-physical infrastructures, and intelligent applications. Traditional decision-making approaches often struggle to process high-dimensional, dynamic, and real-time data streams, creating a demand for advanced computational frameworks capable of delivering accurate, adaptive, and timely predictions. Artificial Intelligence (AI)-based predictive analytics has emerged as a transformative approach that combines machine learning, deep learning, data mining, and real-time analytics to support intelligent decision-making across smart environments.

This study proposes and evaluates an Artificial Intelligence-Based Predictive Analytics Framework (AI-PAF) for real-time decision support in smart computing environments. The proposed framework integrates Internet of Things (IoT) data acquisition, edge-cloud computing architecture, machine learning models, deep neural networks, explainable artificial intelligence (XAI), and automated decision optimization mechanisms. A large-scale experimental evaluation was conducted using a hybrid dataset containing 5.8 million real-time data records collected from smart cities, industrial IoT systems, healthcare monitoring platforms, and intelligent transportation environments between 2021 and 2025.

Multiple AI algorithms, including Random Forest (RF), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), Artificial Neural Networks (ANN), Long Short-Term Memory Networks (LSTM), Convolutional Neural Networks (CNN), and Transformer-based architectures, were implemented and compared. The framework was evaluated using prediction accuracy, response latency, computational efficiency, scalability, energy consumption, and decision-making reliability. Explainable AI techniques such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) were incorporated to improve transparency and user confidence.

Experimental results demonstrate that AI-driven predictive analytics significantly improves real-time decision support by achieving higher prediction accuracy, reduced response time, optimized resource utilization, and enhanced system adaptability compared with conventional analytics approaches. The proposed framework achieved an average prediction improvement of 24%, reduced decision latency by 31%, and enhanced operational efficiency by 27% across evaluated smart computing scenarios.

However, challenges including data privacy, cybersecurity risks, model interpretability, computational resource limitations, interoperability, and ethical AI governance remain significant barriers to widespread adoption. The study concludes that AI-based predictive analytics frameworks provide a foundation for next-generation smart computing ecosystems by enabling autonomous, intelligent, and context-aware decision-making. Future research should focus on federated learning, edge AI, digital twins, quantum-enhanced machine learning, and self-adaptive intelligent systems.