Design of an Iterative Dual Metaheuristic VARMAx Model Enhancing Efficiency of Time Series Predictions

Traditional methods, while effective to a degree, often grapple with limitations such as reduced accuracy, specificity, and higher prediction delays, particularly in high-stakes fields like stock market analysis, forest cover monitoring, and medical disease prediction. To address these shortcomings, this study introduces an innovative approach: the design of an iterative dual metaheuristic VARMAx (Vector Autoregressive Moving Average with exogenous inputs) model. This model leverages the robustness of Grey Wolf Optimizer (GWO) to enhance the Vector Autoregressive (VAR) component and employs the Coot Optimization Algorithm to optimize the Moving Average (MA) aspect of the VARMAx process. The rationale behind selecting these specific metaheuristic algorithms lies in their proven efficiency in exploring and exploiting the solution space, which is crucial for achieving accurate time series predictions. The proposed model demonstrates a significant leap in performance metrics when tested across various datasets encompassing stock market trends, forest cover changes, and medical disease forecasts. Compared to existing methodologies, the iterative dual metaheuristic VARMAx model exhibits a 2.9% increase in prediction precision when compared with Fuzzy C Means (FCM), and Empirical Mode Decomposition (EMD) Methods, a 3.5% rise in accuracy when compared with Variational Model Decomposition (VMD) & EMD Methods, a 2.5% improvement in recall when compared with VMD & FCM, a 3.2% enhancement in the Area Under the Curve (AUC), and a notable 3.4% boost in specificity when compared with VMD, FCM & EMD Methods. Moreover, it significantly reduces the delay in predictions by 4.9%. These metrics not only underscore the model’s efficacy but also highlight its potential in delivering timely and reliable forecasts, which are critical in decision-making processes. The models ability also verified with two statistical tests called Model confidence set and Diebold Mariano test procedures at 5% Level of Significance. These tests aim to assess the performance of the VARMAx model compared to competing models across multiple datasets & samples. The impact of this work extends beyond mere numerical improvements; it sets a new benchmark in the field of predictive analytics. The integration of metaheuristic algorithms with the VARMAx model marks a paradigm shift in addressing the challenges posed by time series data samples. This model’s adaptability and precision make it a valuable tool for researchers and practitioners across various sectors, paving the way for more informed and effective strategies in areas ranging from financial forecasting to environmental conservation and healthcare management. The outcomes of this study, therefore, hold substantial implications for both academic research and practical applications, contributing significantly to the advancement of time series prediction methodologies.