Project Leads: Nidhil Nayudu, Jake Gwinn
As daunting as the stock market may seem, it typically follows certain trends. Many of us have investments in the market, or will eventually, so it is important (and interesting) to understand what could influence stock prices. Using past market data, we can highlight the important factors that do just that. Then using that information, we can build a model to predict stock prices!
Initially, we will decide on a company to analyze and get their stock history and news headlines from Yahoo Finance and Kaggle. Then, we will guide you through cleaning this data and performing exploratory data analysis (EDA) to reveal some foundational relationships in the market. Now that you have the fundamental skills to do this yourself, we’ll split into small groups and go through this procedure for various companies in the market. Then using the important features from our EDA, we will walk you through the basis of LSTM neural networks and how to design your own for predicting stock prices. You will also be able to work with sentiment analysis with news headlines to update our model, if needed. You’ll each then be able to show off your work on a website we will create!
- Search for datasets relevant to our design statement
- Perform EDA on various metrics of financial data using - - Python’s NumPy, Pandas, MatPlotLib, etc libraries and basic sentiment analysis on news headlines using Transformers
- Learn basic neural network structures and various types of layers through TensorFlow
- Design, train, and analyze ML model to predict stock price trends
- Present findings and work through web-app using Streamlit (Optional)