Skip to content
OtisMurrayPublic

About

Here is a collection of my major projects developed through my college career.

Resources

Stars

1 star

Watchers

0 watching

Forks

Repository files navigation

Otis Murray Portfolio

This repository is a collection of data science, machine learning, analytics, and software projects I built through coursework, research, hackathons, and team collaborations. The work spans predictive modeling, environmental forecasting, text classification, clustering, exploratory analysis, and database design across both R and Python workflows.

Portfolio Highlights

An end-to-end wildfire risk forecasting project that combines environmental data engineering, machine learning, and application development.

  • Built for the GDG @ Penn State Solution Challenge Hackathon: Innovate for Impact on April 11-12, 2026
  • Built forecasting pipelines in Python using weather data, NASA FIRMS fire observations, and gridded spatial features
  • Trained and compared tree-based models including Extra Trees and XGBoost
  • Exposed predictions through a FastAPI backend
  • Built an interactive React and Leaflet frontend for mapping wildfire risk and active fire detections
  • Designed as an impact-oriented climate and resilience project aligned with the Google Solution Challenge format

A large-scale environmental data processing and forecasting project focused on PM2.5 air pollution over North America.

  • Processed NetCDF air quality data into cleaned modeling datasets
  • Built forecasting workflows with CatBoost, XGBoost, and linear regression approaches
  • Generated figures and supporting analysis scripts for model evaluation and communication

A text classification project in R that predicts Reddit post categories using embeddings and gradient-boosted models.

  • Classified posts into 11 categories
  • Performed preprocessing, dimensionality reduction, visualization, and hyperparameter tuning
  • Logged model experiments to track tuning decisions and performance across runs

A regression modeling project that predicts IC50 values for the Omicron variant using patient and covariate data.

  • Used R for preprocessing, feature engineering, dimensionality reduction, and modeling
  • Applied XGBoost to a biomedical prediction task
  • Saved tuning attempts and model artifacts for reproducibility

An unsupervised learning project in R focused on breed pattern discovery and probabilistic clustering.

  • Normalized features with caret
  • Applied PCA for dimensionality reduction
  • Used Gaussian Mixture Models with ClusterR to assign probabilistic breed group memberships

A structured prediction project in R for estimating house sale prices from training and testing data.

  • Cleaned and reformatted raw housing data
  • Created group-based price estimates using house quality and condition labels
  • Organized work into reproducible src, data, and models folders

A team database project designed to improve volunteer management for the United Helpers Organization.

  • Helped design a relational database for volunteers, tasks, assignments, packages, and items
  • Wrote SQL for table creation, data integrity rules, and operational queries
  • Prepared the system for web access through PostgREST support

A Python-based exploratory data analysis project studying how occupation, demographics, and lifestyle habits relate to sleep quality and duration.

  • Used Pandas and Matplotlib for cleaning, analysis, and visualization
  • Investigated behavioral and demographic patterns tied to sleep outcomes

A set of hands-on notebooks covering practical machine learning workflows and experimentation.

  • Data preprocessing and white-box model training
  • Random forest modeling for mental health risk analysis
  • Customer segmentation
  • Classification projects on image and tabular datasets

Additional Collaborative Work

Collaborative Penn State Hackathon Fall 2024 project that won Best Use of MongoDB Atlas.

  • Built a journaling application designed to flag warning signs of mental illness from user-written entries
  • Worked on an applied AI-centered team project with product, data, and software components

Skills Demonstrated Across This Portfolio

Machine Learning and Statistics

  • Regression and classification
  • XGBoost, CatBoost, Extra Trees, Random Forests, Gaussian Mixture Models
  • PCA and other dimensionality reduction workflows
  • Hyperparameter tuning, model comparison, and evaluation

Data Engineering and Preparation

  • Feature engineering
  • Missing value handling, scaling, encoding, and data cleaning
  • Working with CSV, NetCDF, JSON, and spatially gridded datasets
  • Organizing raw, interim, and processed data for reproducibility

Software and Visualization

  • R, Python, SQL, FastAPI, React, Leaflet, and PostgREST
  • Pandas, Matplotlib, caret, ggplot2, xgboost, ClusterR, and scikit-learn workflows
  • Interactive map-based visualization and model output communication

Collaboration

  • Individual coursework projects
  • Research-style modeling and analysis
  • Team software/database builds
  • Hackathon development under time constraints

Repository Structure

Each top-level folder represents a separate project. Some projects are full application or modeling repositories, while others are focused notebooks or academic deliverables. Together they show a range of work across:

  • Machine learning
  • forecasting
  • environmental analytics
  • text modeling
  • SQL and database systems
  • exploratory data analysis

About Me

I am interested in statistical modeling, data science, machine learning, and applied analytics. This portfolio reflects the way I approach projects: start with the data carefully, build interpretable and well-structured workflows, and turn technical results into something useful and understandable.

About

Here is a collection of my major projects developed through my college career.

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages