A Prediction Algorithm For Unlabelled Tweets for Stock Market Data

This paper presents a comprehensive methodology for sentiment analysis and stock market prediction leveraging Twitter data and machine learning techniques. The proposed approach integrates data collection from Twitter, clustering using the K-means algorithm, sentiment labelling with a rule-based Mamdani fuzzy system, and classification using various classifiers. The experimental results, based on the evaluation of precision, recall, and F-measure, showcase the efficacy of the methodology. Specifically, the Deep Neural Network (DNN) with two layers emerges as the top-performing classifier, exhibiting impressive average precision, recall, and F-measure values of $0.977,0.921$, and 0.948, respectively. In comparison, while the K-nearest neighbors (KNN), Decision Tree (DT), and Naive Bayes classifiers also demonstrate competitive performance, they fall slightly short of the precision, recall, and F-measure achieved by the proposed DNN model. These findings underscore the superiority of deep learning techniques in accurately capturing sentiment from Twitter data and predicting stock market trends. Overall, this paper contributes valuable insights into sentiment analysis and stock market prediction, offering practical methodologies that can inform investment decisions and enhance financial market analysis in today’s digital age.