• Built a predictive model (XGBoost) to score the risk of Catalytic Converter loss claim in AWS, applied it on more than 3 million policies. The Top1% group scored by model is 80 times risker than average. • Wrangled 10GB+ data in Snowflake and AWS, analyzed and visualized it in both time series and geographic format in AWS Sagemaker. • Presented the findings and propositions to VP-level stakeholders in product, pricing, and claim teams, highly praised for the results and communication skills.

Used the company’s analytical tools (similar to Tableau) to do data wrangling, text preprocessing, and data visualization of various types of data including social buzz, the share of voice, and consumer reviews.

1. Metrics Building: Built with Excel a scoring system on the subject regions’ water security, which was promoted as a blockbuster product of MyH2O and received recognition from the Director of Water Bureau of Wenzhou City.2. Data Analysis: Cooperated with teammates to employ Tableau to visualize 5,000+ data from 2,000+ questionnaires;3. Communication and Research: Wrote the first Case Studies Report of MyH2O. Communicated with 21 partners, 6 undergraduate teams, and 3 donors, and finally sorted out a complete review of 6 projects in the past five years.