Projects - Hi, I'm Sebastian Marrero

Built fraud detection models using Logistic Regression and Random Forest (scikit-learn)
Implemented SMOTE resampling to address severe class imbalance (imbalanced-learn)
Performed exploratory analysis on application velocity, credit risk, email domain, and behavioral patterns (pandas, seaborn)
Engineered features and imputed missing values using NumPy and pandas for robust model performance
Visualized class imbalance, fraud likelihood by categorical attributes, and feature impacts (matplotlib)
Identified key fraud signals such as high velocity metrics and weak credit histories using feature importance rankings
Deployed notebook-based reporting with Jupyter and integrated all outputs into a Jekyll-based portfolio site

Built churn prediction models using Logistic Regression and Random Forest (scikit-learn)
Performed exploratory analysis on tenure, support behavior, subscription type, and user engagement
Visualized churn patterns and feature impacts using matplotlib and seaborn
Improved model AUC through Random Forest tuning and feature scaling with StandardScaler
Delivered business insights to guide customer retention strategies based on high-risk behaviors

Analyzed global start-up trends using SQL (MySQL Workbench) and Python (pandas, SQLAlchemy)
Explored investment, funding, and valuation metrics across industries, countries, and investors
Created business-driven visualizations and ROI metrics using Tableau Public

Analyzed NYC short-term rental trends with SQL + Python from Airbnb housing data
Performed Sentiment Analysis using TextBlob in Jupyter Notebook
Integrated and enhanced multiple datasets using Python and PostgreSQL to uncover time-based and categorical trends