Study on AI-Assisted Statistical Approach for Improving Stock Price Prediction Accuracy

Abstract

Stock price (SP) prediction is crucial for financial decision-making, yet achieving excessive accuracy remains challenging due to market volatility. Current models frequently struggle with capturing the complexities of SP fluctuations, leading to significant prediction errors. This study aims to improve SP prediction accuracy through a unique technique that uses AI-assisted statistical techniques with the Redefined Spotted Hyena great-tuned Dynamic Gated Recurrent Unit (RSHDGRU). The dataset includes the closing costs of numerous stocks influenced through market demand, corporate performance, and economic situations. Pre-processing using Z-score normalization to standardize the statistics. The proposed RSH-DGRU model significantly outperforms traditional techniques, achieving a R-squared (R²) value of 0.9852, a Mean Absolute Error (MAE) of 15.624 and Root Mean Square Error (RMSE) of 20.321. These results reveal the effectiveness of the RSH-DGRU in minimizing prediction errors and accurately capturing the complexities of SP fluctuations. By evaluating its overall performance with present fashions, the RSH-DGRU technique showcases stronger predictive capabilities. Financial analysts and investors that have access to a strong instrument for more precise market projections make better-informed investment selections.