Stock Price Prediction Using Recurrent Neural Network Architecture

The financial market is a concept where financial commodities such as golds, silver, valuables and other essential commodities are traded in terms of proceedings between buyers and stockist. Due to the present condition of the stock market, machine learning and deep learning algorithms have been implemented to help buyers and sellers to predict the possible outcome of the financial market. The stock market is sensitized to the political economy atmosphere. However, both types of information are too complex and unstable to gather. The above information that cannot be included in features is considered as noise. This paper presents a Recurrent Neural Network Architecture in predicting stock prices. The dataset used in this study consists of stock price starting from the year 2011-2019; and contains 8 columns. The dataset was preprocessed by transforming them into understandable format, and also removing inconsistence and thereafter scaled and normalized to (0,1) values by creating a scaler object using MinMaxScaler (feature_range (0,1)). We transformed the data to matrix format with rows and columns and reshaped the data to arrays before transforming. The data was being reshaped to a 3D array in other for the Long-Term Short Memory Algorithm to read and train the model. After building and training our Recurrent Neural Network model, we downloaded a test dataset, append it to our trained model and made prediction, we plotted a line graph of the actual stock price and the predicted stock price against time. The plotted graphs shows the original profits made by the company during a stock exchange versus the predicted profits the company will have in future.