Generating Stock Market Data and Making Predictions Using GAN and Neural Networks

Solving time series forecasting has always been a problematic task for computer scientists with always uncertainty among the stakeholders regarding the behaviour of the time series data. With the help of machine learning and artificial intelligence, computer scientists have provided a solution by developing deep learning models which can solve time series forecasting problems. Stock market prediction is an example of time series forecasting and deep learning models can predict the stock market and help the shareholders in making a better decision of either buying, selling, or holding the share. In recent years, researchers have found that deep learning models have been showing great results and are outperforming traditional machine learning models. Hence, in our study, we decided to use multi-layered neural networks. We will be using the recurrent neural network (RNN), the long-short term memory neural network (LSTM), and the gated recurrent units neural network (GRU) during the course of our study. In our study, we will use these neural networks to predict the closing price of a stock market share. Additionally, for the stock market data, we will also use a generative adversarial network (GAN) to generate fake data. We will then use this fake-generated data to make predictions using the above-mentioned neural networks. We will then evaluate these models on different evaluation metrics and critically analyse the predictions made by our neural networks.