Stock prediction using RNN (LSTM and BiLSTM)

In this approach, the goal was to predict stock prices using a Recurrent Neural Network, RNN. For this purpose, several model architectures for stock price forecasting were presented. Also, tried to test different model parameters such as number of layers, nodes and drop outs and fit over several batch sizes, epoch sizes, optimizers and different activation functions to investigate the effect of different parameters in accuracy of the model. It is clear that, With proper layers and hyper parameters, the LSTM and the Bi-LSTM models give good results for stock prediction.