Sentiment Analysis of Top Stock Market Companies

Abstract

In the current era, data grew in volume, variety, and velocity; therefore, handing the term big data. While conventional data type consists of formatted data, big data involve structured, semi structured and unstructured data for which enhanced tool is requisite for processing the data to pull out the value-added information. The characteristics of big data are briefly discussed in this paper particularly, the 5 Vs of big data which include Volume, Velocity, Variety, Veracity and Value as features used for the selection of datasets for analysis.
A set of 3 million tweets associated with five popular international enterprises, including Amazon, Apple, Google, Microsoft and Tesla was chosen to illustrate the characteristics of big data. This data set is both numeric and non-numeric; numeric data includes only tweet ID and metadata, while the non-numeric data includes the text of the tweet from 2015 to 2020. The study used this dataset to assess its applicability to the 5Vs framework and perform the descriptive statistical analysis to determine broader trends using Python.
The result of this study shows that the sentiment of all companies mainly ‎neutral sentiments, ‎followed by positive sentiments; a lesser quantity of tweets expressed negative ‎‎sentiments. Through machine learning approaches, the best result was achieved by the SVM model with the correct rate of sentiment prediction at 96%, which is better than the other models like LR and Decision Tree. This analysis highlights the significance of real-time sentiment monitoring and spotlighting prescriptive recommendations, such as creating content strategies that actually engage audiences by successfully driving their interactions. By implementing these strategies, enterprises can strengthen their positions in the market and control their weaknesses more readily.