Kuan-Chih Lee

Data Scientist

Blend360, LLC

07.2022 - Current

CVS Pharmacy, Exploratory Data Analytics: Improved the IHE campaign reach rate from 29% to 33% by analyzing historical data to determine optimal outreach timing, including the best days and time windows to contact different patient types.
CVS Pharmacy, Predictive Modeling and Optimization: Integrated an XGBoost engagement model to predict patient engagement with a proximity optimization solution to minimize distances and consider constraints in store assignments, resulting in tailored prospect lists for each store to enhance outreach efficiency.
Sunovion Pharmaceuticals, Marketing Mix Modeling: Enhanced client campaign strategies by employing Marketing Mix Modeling to elucidate the impact of each marketing channel for various groups and to determine response scores.
Sunovion Pharmaceuticals, Prediction Modeling: Facilitated campaign planning by delivering prospect lists to the client, employing K-means clustering on healthcare professionals based on engagement propensity and response scores to categorize them into different levels.
LinkedIn, A/B Testing: Performed incrementality testing by executing A/A validation and A/B tests at a DMA-level granularity, assessing the efficacy of advertisements on key performance indicators like weekly active users and conversions.
Benefytt Technologies, Machine Learning Modeling: Enhanced annual enrollment campaign response rate from 0.42% to 0.64%, reducing cost per acquisition to $800 by employing ensemble models (XGBoost, Logistic Regression, LGBM) across 90 million universes.
Benefytt Technologies, Data-Driven Marketing Tactics: Developed targeting strategies prioritizing the most valuable user segments generated from K-modes Clustering in different angles, such as demographic, psychographic, behavioral, and geographic, and translated them into meaningful stories to drive strategy, extending our company's contract with the client.
Benefytt Technologies, Data Visualization: Visualized SHAP values to showcase feature contributions to responders and conducted ad hoc analyses, tracking response and conversion rates over time, delivering concise visual insights to stakeholders.
Benefytt Technologies, Data Wrangling: Constructed a database using SQLAlchemy, optimized the pipeline to ingest diverse third-party and client data from S3 bucket, Snowflake, and shared drives, along with outputs from Alteryx, directing the final data into Databricks for modeling purposes.
Blend360, Recommender System Engineering: Engineered an advanced recommender system encompassing item ranking and collaborative filtering algorithms, which was subsequently encapsulated into a versatile Python solution for extensive company-wide application, with a particular emphasis on enhancing grocery product recommendations.

Similar Profiles