Sentimental Analysis of Tweets using Machine Learning

The stock market forecast is a long-term lucrative subject for researchers from various fields. An accurate market prediction is of great importance to investors, but stocks are driven by unpredictable factors, such as microblogs and media, making it impossible to forecast the stocks index based on historical data alone. It is understood that the financial market is knowledge responsive, stock prices represent all the existing knowledge, and price changes can be in response to news or events. For many scholars and economists, the practice of estimating stock markets was a challenging task. Several types of research have been performed in particular to forecast stock market behaviour using algorithms of machine learning. In this paper, we use media platforms and business news data to examine the impact of these data on the accuracy and consistency of stock market forecasts. To accomplish our objective, we used the twitter API and processed it for further analysis. Two machine learning methods were used to evaluate tweet sentiment, namely Naïve Bayes classification and Support vector machines. By comparing each model, we find that the support vector machine provides higher accuracy by cross-validation.