RNN to predict Google Stock Prices

A Recurrent Neural Network coded in Python using Tensorflow's Keras to try to predict Google Stock prices.

Installation

• Download and install Python3 from here

• I recommend using virtualenv to create a Python environment. It's good practise to create and use different environments for different projects. Open a terminal and type:

  pip install virtualenv

• Create a Virtual environment with some name say rnnenv.

  • Windows

  virtualenv rnnenv
  cd rnnenv/Scripts
  activate

  • Linux:

  source rnnenv/bin/activate

• Clone this repository, extract it if you downloaded a .zip or .tar file and cd into the cloned repository.

  • For Example:

  cd A:\RNN-to-predict-Google-Stock-Prices-master

• Install the required packages by typing:

  pip install tensorflow==1.15
  pip install pandas
  pip install matplotlib
  pip install numpy
  pip install scipy
  pip install -U scikit-learn

Running

python rnn.py

Dataset

The dataset contains high, low, close and open values of Google's stock prices, date and the volume. I tried to predict the open values based on past open values

Architecture

The architecture of the RNN is 4 sets of LSTM (Long short term memory) layers and Dropout regularization layers with the LSTM and Dropout layers sandwitched between each other. The number of time steps chosen for the RNN is 60. The model is trained on 100 epochs with a batch size of 32. The optimizer used is adam and the loss function is mean squared error.

Results

After successful prediction, the results were visualized using the matplotlib library.