Analyzing Market Trends Based on Social Media Data

Nowadays, people share their everyday life issues, happy moments, ideas, plans via Internet on social media platforms and read others' posts. This allows an exchange of ideas and a quick response to social cases. All of these posts create a huge amount of digital data which can be analyzed.

Other than their immediate effect, these social media posts sometimes influence some social indicators in the future, economy and markets, for example. In this project, we have decided to analyze the correlation between cryptocurrency markets and people's thoughts as represented in their posts on the social media platforms. For example, by creating a robust machine learning model, we can predict the future price changes in the cryptocurrency markets and decide whether to invest in them or not.

Surely, to understand people's thoughts are tough and complex for computers to understand but machine learning and neural networks might enable us to classify people's thoughts by looking at the words in their tweets. What kind of tweets cause bullish or bearish cryptocurrency markets tomorrow? That question is the foundation for that project.

Not only tweets, but also Google search frequencies, Euro/USD parity, XAU/USD parity, and some stock market indices affect the cryptocurrency markets. We will also take their effect in the cryptocurrency markets.

We also use, other datasets to test the models designed for the project. One such dataset contains 1.5 millions of tweets for sentiment analysis. The other dataset contains 4 millions Amazon reviews.