Analysis and Prediction of Stock Market Using Twitter Sentiment and DNN

Abstract

The analysis of market is one of the important tasks for data analysts. The stock market is very volatile and many models are developed to predict the market by training the model on historical data. But sentimental analysis approach to predict the market is not common. Here we are using REST APIs to capture the twitter data and sentiment analysis using Google’s Natural Language Cloud API along with machine learning techniques like linear regression, Random forest and DNN are used to predict the bear and bullish curves of the stock data. It was found that positive response in the tweets is observed to result in a rise in market value and negative tweets results in fall of the market value. Neural-networks with Feed-forward technique are found to predict prices of stocks with less error in comparison with other used prediction techniques.