Shengyu Wang Email & Phone Number

California, United States

Rochester, Ny, Us

Rochester, New York, United States

• Developed Innovative Data-Assimilation Numerical Models: Designed and developed 3D data-assimilation models that incorporate regression techniques, time-stepping methods, and optimization algorithms to quantify oceanic methane production from large datasets, achieving a 91% R² and enhancing prior estimates by 23%. • Conducted Time-Series Analysis on Climate Data: Performed data validation, calibration, and time-series analysis on predictive climate models, forecasting that methane emissions from the ocean will double by the year 2300.• Modeled Carbon Removal Capability using AI & Deep Learning: Constructed numerical model to evaluate the ocean’s methane absorption capacity using artificial intelligence (neural networks), random forest regression and optimization algorithms, forecasting a ~50% reduction in ocean’s carbon capture and storage ability by year 2100.• Constructed Ocean Physical Transport Model: Developed the first-ever 1D ocean sediments methane transport model by solving systems of ordinary differential equations and employing machine learning algorithms, projecting a ~45% increase in sedimentary methane production by the year 2100.

Rochester, New York, United States

Led a team in processing and analyzing over 592,000 congressional tweets using Natural Language Processing and Word Vectorization. Applied supervised training and ensemble learning for political partisanship prediction, resulting in an 89.3% model accuracy.

Rochester, New York, United States

Processed and analyzed over 166,000 COVID-19 time series observations, implementing different clustering techniques, advanced neural network, and deep learning methods to develop predictive models that achieved accurate COVID-19 forecasts with R2 values above 85%.

Shanghai, China

• Applied X-12-ARIMA and Hodrick-Prescott filter to analyze time series data of Chinese iron ore import prices from 2010 to 2022, identifying significant seasonal instabilities due to COVID-19 and the Russo-Ukrainian War.• Developed a quantile regression model revealing the US Dollar Index and Baltic Dry Index as key determinants of Chinese iron ore import prices, suggesting that increased RMB internationalization will reduce import costs.

Los Angeles, California, United States

Los Angeles, California, United States

Applied Bayesian statistics and Laplace smoothing on two earthquake prediction models, improving model accuracy by 30% and projecting an increase in earthquake frequency in Southern California for the next million years.

Phd Candidate, Geological And Earth Sciences/Geosciences, 3.96 Out Of 4.0

Co-advised by the Mathematics Department

Bachelor Of Arts - Ba, Computational And Applied Mathematics, Geological Sciences

Dual Degrees

Bachelor Of Arts - Ba, Geological And Earth Sciences/Geosciences

High School Diploma