Data Preprocessing for Stock Price Prediction Using LSTM and Sentiment Analysis

Abstract

Stock market marks an intrinsic aspect of a nation’s economy. Being the current buzzword, people are curious to learn how to effectively invest in order to benefit themselves. Right investments have led people to earn enormous profit whereas some had to forfeit. The risk factor in the stock market has always been dreadful for new investors and into the bargains of the experienced ones, but with the evolving technologies it is now trouble-free to make predictions about the stocks. The company’s historic performance succor the investors furthermore different algorithms assist the prediction. In order to extrapolate predictions it becomes indispensable to preprocess the data. In this paper we have made an attempt to model the historic prices of the TCS- Tata Consultancy Services and calculated its accuracy for different epochs and batch sizes, forbye the ramifications of data preprocessing. Further the tweets related to it are scrutinized for the model. Our paper makes an attempt in providing a panorama over different data manipulations and the fidelity procured, we have provided a comparative study herein.