Publication

Abstract : This paper presents a hybrid prediction approach for stock price forecasting by combining Autoregressive Integrated Moving Average (ARIMA) and Seasonal Autoregressive Integrated Moving Average (SARIMA) with Long Short-Term Memory (LSTM) or Gated Recurrent Unit (GRU) models to create ensemble models. The objective is to enhance the precision and dependability of stock price predictions by using the benefits of each individual model. The ensemble models proficiently manage trends, seasonality, non-stationarity, and identify long-term correlations in the TITAN stock price data. This research uses the Volume Weighted Average Price (VWAP) as a key factor to accurately depict the behavior of the market. The ensemble models are structured using the residual method resulting in improved predictions. The superior performance of the ensemble models is confirmed through evaluation using Mean Squared Error (MSE) as the metric. Visualizations in the form of graphs are employed to compare actual and predicted values, demonstrating the accuracy and performance of the models in capturing stock market trends and patterns. The models also exhibit high accuracy in forecasting future stock prices, highlighting their practicality in real-life scenarios.