Deep Learning Paradigm for Time Series Stock Prediction

Abstract

Stocks have become established and widely recognised as a new electronic alternative exchange currency system, with significant consequences for emerging nations and the global economy in general. The ever-expanding financial industry is marked by substantial volatility and sharp price variations over time. Advanced deep learning models have been shown to deliver efficient and trustworthy stock forecasts when applied to highly nonlinear time-series situations. The goal is to create a deep learning model for stocks with superior forecasting accuracy. Their approach aimed at identifying and understanding the factors which influence the value formation of these digital currencies. Time series stock prediction can be time efficient and helps to avoid losses. The data is refreshed every second such that new stock prices will be available. This system reduces the man’s work and gives the information about required stock in seconds and provides all the information in a single place. The user interacts with a chatbot, where the user can get the current price, news, and prediction and determine the feelings of the people regarding the particular stock. The feelings of the people regarding the stock are analyzed using Twitter data. The Prediction is done using Neural Network, KNN, and Linear Regression.