Mining Trading Signals among High Market Capacity Stocks in Nasdaq

For the final project of Data Mining, our team first accessed stock price time series data with more than 5 billion market capacity via MarketStack API. Then, we conducted feature engineering based on trading signals, such as K-line patterns, Moving Average, Relative Strength Index, etc. Lastly, we ranked and demonstrated the predictive power of signals using Decision Tree, T-Test and histograms.

Solar and Wind Plant Day-ahead Power Generation Forecast

During the time I worked as a Data Analyst in Greenwich Commodities LLC, I applied multiple Machine Learning models, and selected the optimal linear regression model. Also, I built and maintained databases for National Weather Service data and PJM power plant generation data, and finally visualized the daily trend of generations using folium and provided insights for future market price.

U.S. Electric Vehicle (EV) Market Dashboard

For the final project of SI 649 - Information Visualization, our team developed a website presenting EV’s geographic distribution and holding rate with Tableau Public. It helped with examine the patterns of EV market share and contribute to the decision-making to potenial EV consumers.

Used Cars Search Engine for Wolverines

With information retrieval techniques, our team applied Beautiful Soup library to scrape used cars and their complaints data from Carfax and NHTSA. Then, we retrieved most relevant results using improved BM25 score and Learning to Rank approaches. Lastly, we utilized TextRank algorithm to extract keywords of car model reviews and displayed them with word clouds.

IMDb Top 1000 Movie Recommendation System

This project accessed IMDb movies and reviews via API Keys, organized the movie dataset into a tree structure, created the system based on the hierarchy of users’ preferences,and lastly presented interactive interface with Flask.