Projects
Here are some data science projects I've completed.
NLP on Amazon Customer Reviews
Team Leader
-
Used PySpark to pre-process over 50GB of Amazon Customer Reviews in distributed cloud storage.
-
Established structured Hive tables (~4.4M rows each) for streamlined data management, retrieval, and efficient filtering.
-
Developed a SparkNLP pipeline for sentiment prediction on customer reviews, achieving 87.9% accuracy on unseen categories through the integration of machine learning models (Naive Bayes, etc.).
Tools used: PySpark, SparkNLP, GCP, Hive
Full slide can be reached at here.
Diabetes Risk Prediction
Team Co-Leader
-
Trained, tuned and ensembled Random Forest, XGBoost, and Light GBM models on DataProc to early detect diabetes with 90% accuracy
-
Developed and deployed a web app prototype for diabetes risk prediction with Github and Streamlit
Tools used: Python, Dataproc, Github
Code repository can be reached at here.
Telco Customer Churn Analysis
Team Leader
-
Built a scalable SQL database for 7K+ customer data; made 5 typical churned customers portraits with SQL queries
-
Created Business Intelligence dashboards with Tableau including 10+ reports on data insights; Generated 5 possible advice to reduce customer churn
Tools used: OpenRefine, MS Excel, Python, MySQL, Tableau
Full slide can be reached at here.
