Forecasting the Direction of Stock Trends Using Machine Learning and Twitter

Authors

  • Rahul Bhatt, Gesu Thakur, Luxmi Sapra

DOI:

https://doi.org/10.17762/msea.v71i4.831

Abstract

As a consequence of technical advancements and the creation of new machine learning models, there has been increased interest in the study of stock market data. This is because these models provide traders and businesspeople with a platform from which to choose more lucrative companies. The information obtained from sources that are considered to be conventional media has had a considerable impact on the movement of stock prices. However, in recent times, platforms for online social networks have been pushing for a strategy that is more efficient in the dissemination of this information. The information that can be found on social networks may be of great use in determining the various perspectives and sentiments that individuals have on certain topics. Because of the volume and variety of these data, an improved machine learning model is one of the options that is continually being investigated for use in daily forecasting. As a direct consequence of this, an exhaustive comparative examination of the stock market models that have previously been put into use has been carried out as part of this effort. The information on Apple stock comes from Yahoo Finance and Kaggle, respectively. The classification technique did not provide sufficient confidence to be used to connect stock movement with feelings expressed on Twitter, with accuracy values ranging from 53 to 56 percent. This prevented the approach from being deployed. Utilizing the regression tactic, as opposed to the classification approach, resulted in superior outcomes. These models mostly depended on previous prices, and as is evident from the patterns, the emotion on Twitter was not a good predictor of changes in stock values. The XG Boost regressor turned out to be the most accurate model when attempting to forecast future prices.

Downloads

Published

2022-09-16

How to Cite

Rahul Bhatt, Gesu Thakur, Luxmi Sapra. (2022). Forecasting the Direction of Stock Trends Using Machine Learning and Twitter. Mathematical Statistician and Engineering Applications, 71(4), 2729–2738. https://doi.org/10.17762/msea.v71i4.831

Issue

Section

Articles