zixuanyang0421

I am a Math and Computer Science student at the University of Michigan with hands-on experience in quantitative research and high-frequency trading data analysis. 

My work spans quantitative research, machine learning, and large-scale data science. At Eastmoney, I processed high-frequency Level-2 market data and built Python/SQL/DolphinDB pipelines for minute-level factor construction, data validation, and factor documentation. I worked with order, trade, snapshot, and minute-bar data and supported model-oriented factor evaluation, including multi-head Transformer-block factor representations with 239 time steps and 186-dimensional features, as well as an XGBoost-based modeling workflow. In my EECS 545 project, I worked on meta-learning and reinforcement learning experiments using Transformer-block representations, rotation Lie group latent features, and DQN/PPO benchmarks, focusing on preprocessing, feature construction, model comparison, and result interpretation. In my data science research, I built DataCite/OpenAlex pipelines to study scientific dataset reuse and novelty, constructing paper–dataset linkages, computing reuse frequency and Rao–Stirling/cosine-similarity novelty measures, and producing statistical analyses and visualizations.