Yandong Zhang | Portfolio

About Me

I'm a data analyst and aspiring business analyst pursuing my M.S. in Business Analytics and Risk Management at Johns Hopkins University (expected July 2025). I specialize in supply chain optimization, predictive modeling, and data visualization, aiming to turn complex data into actionable insights.

Experience

Yonghang Shipping

Data Analyst • Zhejiang, China • Jun 2023 – Sep 2023

Optimized supply chain efficiency by analyzing shipment data trends, reducing potential delays by 15%.
Cleaned and restructured salary and expense datasets using MySQL, improving finance team's data retrieval efficiency by 40%.
Developed a predictive model using Random Forest Regressor to forecast monthly spending, resulting in more efficient budget allocation and reduced over-expenditures.
Designed an interactive Tableau dashboard to visualize spending trends, enabling executives to identify cost-saving opportunities and improve decision-making.

Littlefield Supply Chain Simulation

Johns Hopkins University • Feb 2025 – Mar 2025

Designed dynamic reorder policies and safety stock levels, boosting service level to 98%.
Analyzed lead time variability and capacity constraints, reducing stockouts by 85%.
Tuned production schedule and machine allocation to maximize throughput under budget limits.
Secured 1st place overall by achieving optimal balance between cost, inventory, and service KPIs.

The Purdue University Data4Good Case Competition 2024

6th Place Overall, 3rd in Eastern Division • Sep 2024 – Nov 2024

Addressed the challenge of automating grief journey mapping by implementing and optimizing Random Forest and XGBoost models, achieving high classification accuracy with F1-Score of 0.941.
Fine-tuned the Llama 3.2 (3B) model with customized prompts to better capture nuanced grief-related emotions, significantly improving text-based classification of survey responses.
Led collaboration across a 4-member team, designing an aggregation framework combining traditional models and LLMs, which contributed to the final leaderboard success.
Achieved a score of 945/1000, demonstrating expertise in foundational AI concepts, including the development of AI solutions using Microsoft Azure services.

Kaggle Competition CIBMTR: Equity in post-HCT Survival Predictions

Top 10.6% (Rank 352 / 3,325 Teams) • Dec 2024 – Mar 2025

Performed extensive EDA and correlation analysis to guide feature selection, combining heatmaps and model-driven importance scores to refine inputs from complex synthetic healthcare data.
Built a stacked ensemble integrating a hyperparameter-tuned neural network with XGBoost, CatBoost, and LightGBM models to effectively capture non-linear patterns in survival prediction.
Designed a robust ensemble strategy by blending personal model outputs with top-performing public notebooks, enhancing leaderboard stability and final performance.

Projects

Netflix Content Analysis Report

Data analysis and visualization of the Netflix library.

This project provides an in-depth analysis of the Netflix content library. It uncovers key trends in content types, genre popularity, and release strategies, offering data-driven insights into Netflix's programming decisions.

View Report →