Forecasting the stock price value using Gated Recurrent Units (GRU) Neural Networks model

Abstract

Forecastingstock price movements is a complex challenging task in the financial domain due to inherent volatility and unpredictability of the stock market.In this study we propose a novel approach to predict stock value using Gated Recurrent Units(GRU) Neural Networks, a variant or Recurrent Neural Network(RNN) known for their ability to capture long term dependencies in sequential data.The proposed GRU NN model showcases significant potential in forecasting stock price values, empowering investors, financial analysis and traders with valuable insights for informed-making. Its application can aid in minimizing risks and maximizing of returns in the ever-evolving stock market landscape. The findings of this study contribute to the growing body of research in financial forecasting using machine learning techniquesand provide a strong foundation for future advancements in the domain. The GRU Neural Network Model involves collecting historical stock price data trading volumes and relevant market indicators.After preprocessing the data to handle missing values and outliers,we engineer informative features,including technical indicators and sentiment scores derived for external sources to enrich the models understanding of the market dynamics. The GRU Model is then trained on prepared dataset to learn complex relationships between historical stock price patterns and other features.Then the proposed architecture model is evaluated using accuracy measure obtained from the loss function Mean Absolute Eror(MAE),Mean Absolute Scaled Error(MASE),Accuracy Percent ,Root Mean Squared Error(RMSE),Mean Absolute Percent Error(MAPE) the accuracy measurements represent lower accuracy, true accuracy and higher accuracy in using the model.