mobidian/stock-prediction-using-lstms

Built a stock prediction program in which the user can pick any company available on Yahoo! Finances and be able to predict the closing price of the stock based on the set date with high accuracy.

Stock-Prediction-Using-LSTMs

Built a stock prediction program in which the user can pick any company available on Yahoo! Finances and be able to predict the closing price of the stock with a very high accuracy.

Preprocessing

• Seperation of the data over the last 60 days to use as the testing set and previous values used in the training set.
• Lets user choose the company, start date, and end date for the analytics that they want to predict on

The Neural Network

The Neural Network in use for this model is an Sequential LSTM model.

The model is comprised of:

• Two LSTM Layers
  • Each consisting of 50 Neurons
  • Do not return sequences on the last layer because we want output predicting the next day's value on the dataset.
• Two Dense Layers
  • Conisting of 25 Neurons and 1 Neuron
• Compilation
  • Optimizer: Adam
  • Loss: MSE (Mean Squared Error)

Example: Predicting NVIDIA stock prices with high accuracy

As you can see, the NVIDIA stock price at the end of the cycle starts to become a little unpredictable.

After applying our model and predictions:

The close correlation between the validations and predictions indicate a reliable model for making business decisions.