ARTIFICIAL INTELLIGENCE MODELS FOR PREDICTIVE ANALYTICS USING BIG DATA MINING TECHNIQUES

Abstract

Rapid digital transformation has generated unprecedented volumes of heterogeneous data, creating significant opportunities for predictive analytics while simultaneously increasing challenges related to data quality, scalability, computational complexity, and decision reliability. Conventional predictive models frequently experience performance degradation when processing high-dimensional and continuously evolving Big Data environments. This study aimed to develop and evaluate an integrated Artificial Intelligence framework that combines advanced Big Data mining techniques with hybrid machine learning models to improve predictive accuracy, computational efficiency, and analytical robustness. Quantitative computational research was conducted using large-scale structured and semi-structured datasets processed through data preprocessing, feature engineering, dimensionality reduction, ensemble learning, deep learning, distributed computing, and hyperparameter optimization. Model performance was assessed using accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve, computational time, memory utilization, and scalability. Experimental results demonstrated that the proposed hybrid framework achieved 98.63% prediction accuracy, an AUC-ROC of 0.995, substantially reduced computational time, lower memory consumption, and superior scalability compared with conventional machine learning and deep learning approaches. Statistical analyses confirmed significant performance improvements across all principal evaluation metrics. Findings indicate that integrating intelligent data mining with Artificial Intelligence enhances predictive capability by optimizing the complete analytical pipeline rather than individual algorithms alone, providing a scalable, efficient, and reliable framework for predictive analytics across diverse Big Data application domains.