Kevin Hitt

Using a star schema design to simplify and accelerate data retrieval for complex queries. Consolidating data from the U.S. Bureau of Economic Analysis and Federal Reserve Economic Data, streamlining numerous tables into key categories like CPI and unemployment rates.

Predicting future medical ICD-10 diagnosis codes based on patient visit sequences. It utilized GRU-based sequence-to-sequence and BERT models to generate and predict these codes. While demonstrating potential, further development is required for real-world healthcare implementation.

Building a binary classification model to predict high car insurance risk based on car attributes. Using eight machine learning models, feature importance aggregation, and hyperparameter tuning, the Random Forest model emerged as the best performer in minimizing false negatives. The results support more individualized insurance pricing, helping insurers reduce losses and retain customers.

Analyzing a dataset of New York City Airbnb listings to interactively explore how different features of the rental properties affected price. Using a Shiny app for interactive association rule mining, the analysis aimed to uncover which factors — such as location, listing descriptions, or host behavior — most strongly correlate with price differences.