Stock Price Prediction Using LSTM GRU and BiLSTM

Abstract

This research investigates the comparative performance of Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bidirectional Long Short-Term Memory (BiLSTM) models for stock price prediction using Tesla dataset. The primary objective was to evaluate the predictive accuracy of these models by minimizing the root mean square error (RMSE) through rigorous hyperparameter tuning. Parameters like number of units (50 and 100), number of layers (1 and 2), L2 regularization (0.001 and 0.01), dropout rate, and batch size was applied to optimize each model’s architecture. This research revealed that GRU consistently achieved the lowest RMSE outperforming both LSTM and BiLSTM with hyperparameters batch size of 32, dropout rate of 0.2, L2 regularization of 0.001, one layer, and 50 units.