An ML-Powered Predictive Analytics Framework for Stock Market Forecasting using Time-Series Analysis Data

Background: The field of stock price prediction holds vital importance because it delivers substantial advantages to the entire population from both private citizens and official institutions. Big data analysis linked with machine learning techniques improves predictive insights through the processing of extensive datasets. Methods: The research implements three deep learning models namely Long ShortTerm Memory (LSTM), Gated Recurrent Units (GRU), and Recurrent Neural Networks (RNN) to forecast Amazon stock prices. The data preprocessing step includes Min-Max normalisation and results in training and testing split sets. Evaluation of performance occurs through R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and RMSE and MAE measurement methods. Results: The stock price prediction results from LSTM provide greater accuracy compared to GRU and RNN with R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>-score 99.61% and RMSE 20.46 and MAE 14.97. Conclusion: The research determines that LSTM networks achieve superior accuracy and reliability for Amazon stock prediction when comparing to alternative advanced deep learning methods.