Yinuo (Nora) Yin
New York, NY · yynlelenora@gmail.com · linkedin.com/in/yynlelenora · yinuoyin.com
Work experience
Two Sigma Investments — Real Estate
New York, NYVice President, Data ScientistJan 2024 – Present
Data ScientistJul 2021 – Dec 2023
- Drove industrial sector investment decisions: beta regressions on long panel data forecasting market and market × size occupancy, blended with rent momentum into a market-level revenue ranking; and lease-level survival models forecasting time-to-lease-up across a ~525K-property national industrial universe. Earlier versions used gradient-boosted decision tree algorithms and synthetic training sets from hedonic modeling.
- Led single-family portfolio construction: causal-inference matching (Coarsened Exact Matching) chained into a Case-Shiller-style repeat-measures growth index, an ensemble long-term price-forecasting model, and automated property-level valuation, combined into a systematic, portfolio-scale selection framework driven by zip-level yield and internal rate of return, spanning ~70M single-family transaction records, used directly in build-to-rent and homebuilder deal underwriting.
- Designed a hypothesis-testing framework evaluating whether alternative datasets carry investment signal beyond traditional real estate fundamentals: matched maritime shipment records, consumer credit-card panel spend, and labor-cost data to individual properties and leases via custom entity-resolution and geographic-proxy joins, with the strongest signals productionized as ongoing inputs to investment decisions and investment committee memos.
- Engineered advanced geospatial features, including point-of-interest and radius/drive-time-based accessibility metrics, improving forecast model performance by 10%+ across multiple industrial models.
- Built a testing framework validating property- and market-level signals against same-store net operating income (NOI) growth for public real estate investment trusts (REITs) across five property sectors; became the basis of an ongoing production REIT analytics effort.
- Scaled team tooling and talent: built a Dynamic-Time-Warping market-matching tool and shared geospatial utilities adopted across multiple deal evaluations, created an LLM-powered research harness that automated experiment workflows, and mentored new hires on the shared codebase and LLM-assisted development.
- Communicated technical work across audiences: presented modeling results and recommendations to investment committees and cross-functional stakeholders, and supported materials for external fundraising efforts.
Urbint (acquired by Itron)
New York, NYSenior Data ScientistApr 2021 – Jun 2021
Data ScientistDec 2018 – Apr 2021
- Led rare-event classification modeling to predict excavation damage risk, capturing 20% of incidents within the top 1% and 60% within the top 10% of highest-risk projects.
- Built and deployed time series forecasting models achieving 80%+ accuracy across service regions to optimize workforce planning.
- Architected reusable ML and geospatial pipelines (H3 hexagonal spatial indexing, weather integration, k-nearest neighbor (KNN) spatial joins) to standardize model development and improve scalability.
- Partnered with Product and Engineering to define KPIs and translate modeling insights into client-facing dashboards.
Fox Chase Cancer Center
Philadelphia, PAGeospatial Data AnalystJun 2018 – Nov 2018
- Applied PCA, hierarchical clustering, LASSO, and elastic net regression to analyze social determinants of advanced prostate cancer outcomes.
- Implemented Bayesian and Cox proportional hazards models on high-dimensional oncology datasets.
- Improved computational efficiency of large-scale analyses using cluster computing.
Education
University of Pennsylvania
Philadelphia, PAMaster of Urban Spatial Analytics · GPA 3.9 · Merit-based ScholarshipAug 2017 – May 2018
University of Wisconsin – Madison
Madison, WIB.S. Landscape Architecture · GPA 3.9 · University Olmsted Scholar; Graduate with DistinctionAug 2013 – May 2017
Publications and talks
- Handorf E, Yin Y, Slifker M, Lynch S. Variable selection in social-environmental data: sparse regression and tree ensemble machine learning approaches. BMC Medical Research Methodology, 20, 302 (2020).
- Mohan K, Yin Y. Developing fine-grained geospatial units that reflect the built and natural world. Spatial Data Science Conference, 2022.
Skills
- Programming
- Python, R, SQL, JavaScript, HTML/CSS
- Modeling & analytics
- Time series and cross-sectional forecasting, regression, classification, survival analysis, causal inference, tree-based models, neural networks, alternative-data feature engineering, geospatial feature engineering, spatial clustering, signal research and backtesting
- Infrastructure & tools
- Git, GCP, Docker, Kubernetes, Airflow, Linux/Unix, ArcGIS, LLM-assisted research tooling