An application that uses neural networks to accurately predict closing stock prices.
Train a PyBrain recurrent neural network to identify patterns in a stock's daily opening, closing, high, and low prices with the goal of predicting closing prices. Remarkable error percentage of only 0.007%.
Develop a GUI in Python using Tkinter, Matplotlib, and Pandas to allow user to input desired stock symbol and to see a graph illustrating it's closing prices and predicted closing price. Threading was used so as to provide an optimal interface experience; running the GUI in the main thread and all other logical and networking jobs in different threads.
Used Yahoo Finance API to request stock information in JSON and fed chronologically to RNN.
Use genetic algorithm to evolve the neural network using low error percentage as an NP optimization solution.
Incorporate sentiment analysis as a variable in the training process.