Temporal data mining techniques for predictive analytics systems
Abstract
Many researchers have identified the extraction of information and knowledge from large databases as a critical research domain within database systems and predictive analytics. Likewise, numerous industrial sectors recognize data mining as a strategically important area with significant revenue-generating potential. In this review study, we explore how data mining facilitates the discovery of meaningful patterns and insights from vast datasets. The field generally comprises six core components: association, classification, clustering, prediction, regression, and outlier detection. The application of data mining techniques is essential for enhancing business opportunities, improving service quality, and understanding user behavior across various information-driven services, including data warehousing and web-based platforms. The primary objective of this research is to review the major techniques employed in the data mining domain. In addition, we conduct an empirical analysis using the benchmark NIFTY 50 dataset from the National Stock Exchange (NSE) of India. Several predictive analytics techniques are applied to this dataset to evaluate their performance. The findings indicate that Long Short-Term Memory (LSTM) networks outperform traditional approaches such as Hidden Markov Models (HMMs).