Colin Middleton

• colindmiddleton@gmail.com
• (360)-206-7812
• middlec000.github.io
• Bellingham, WA

Experienced Data Scientist with expertise in predictive modeling, statistical analysis, and machine learning applications. Proven track record of optimizing business operations through data-driven solutions and translating complex technical concepts into actionable business insights.

Professional Experience

Data Scientist II | Pearl Health | New York, NY June 2024 — February 2025

• Architected and implemented a CFR-compliant pricing framework for Medicare Shared Savings Program clients, resulting in 15% improved forecasting accuracy and enhanced client satisfaction.
• Spearheaded the development of an enterprise-grade data ingestion and versioning pipeline, enabling robust data testing, cross-version analysis, and transparent forecast reporting.
• Designed and deployed a comprehensive tracking system for pricing tool inputs, delivering critical insights for sales funnel optimization and strategic decision-making.
• Led 50+ individualized analyses for strategic prospects, directly contributing to a 30% increase in client acquisition.

Data Scientist I | Pearl Health | New York, NY May 2023 — June 2024

• Developed advanced forecasting models for client and prospect KPIs, leveraging statistical techniques including exponential smoothing, multivariate regression, and Theta modeling.
• Led the comprehensive refactoring of a mission-critical pricing tool (12k+ lines of Python), implementing modular architecture, configuration management, and comprehensive documentation.
• Engineered a proprietary I/O system that increased input accuracy by 25% and reduced processing time by 40%.
• Conducted large-scale statistical analysis of KPI variability across healthcare provider networks, generating actionable insights for healthcare operations nationwide.

Data Scientist | Gordian Data | Seattle, WA July 2022 — May 2023

• Migrated and enhanced a client reporting tool from SAS to Python, improving maintainability and reducing runtime by 10x.
• Resolved a bug in a client pricing tool, increasing pricing accuracy by 5%.
• Contributed to hiring efforts, including designing and implementing an applicant tracking system, participating in job fairs, interviewing candidates, and managing the candidate pipeline.

Education

MS Applied Math | Eastern Washington University | Cheney, WA September 2019 — June 2021

• GPA: 4.0
• Outstanding Graduate (2021)
• Graduate Service Appointment (2020 - 2021)

BS Math | Western Washington University | Bellingham, WA September 2013 — December 2017

• GPA: 3.5
• Chemistry Minor
• Honors Program
• Math Fellow (2015 - 2017)
• Presidential Scholarship (2013)

Research and Publications

Homelessness Prediction | Spokane Predictive Analytics | Spokane, WA June 2020 — October 2023

• Middleton, Colin D., et al. “The value of utility payment history in predicting first-time homelessness.” Plos one 18.10 (2023): e0292305..
• Collaborated with data providers to refine a dataset and presented findings to the Spokane City Council.
• Evaluated predictive models (Linear Regression, Cox Proportional Hazards, LSTM, Logistic Regression, ANN), achieving a peak AUC of 0.81 with Logistic Regression—comparable to research using richer, less operational data.
• Addressed extreme class imbalance (0.39% positive class) via minority oversampling.

Project Experience

The Wordler | Personal Project | Leavenworth, WA March 2022 — April 2022

• Collected data from the Wordle website source code and Kaggle.
• Developed a six-component data filter using regular expressions in Python.
• Displayed words sorted by user-specified criteria: word frequency, letter frequency, and letter position frequency.

Document Clustering | Big Data Analytics Class Project | Spokane, WA January 2021 — March 2021

• Established a data processing pipeline to transform academic papers into Term Frequency-Inverse Document Frequency vectors.
• Assisted with development of a customized K-Means algorithm optimized for sparse, largely non-overlapping vectors.

Predicting Traffic | Whatcom Council of Governments | Bellingham, WA June 2017 — December 2017

• Predicted US/Canadian border crossing traffic using historical and upstream data. A custom model leveraging upstream sensor ratios achieved the lowest error rate (10.83 cars/5 min).

Technical Skills

• Programming Languages: Python (pandas, NumPy, statsmodels, scikit-learn, matplotlib, seaborn), R, SQL
• Data Science Tools: Git, Jupyter Notebooks, dbt, Anaconda
• Machine Learning: Linear/Logistic Regression, Cox Proportional Hazards, Decision Trees, K-Means/Hierarchical Clustering, PCA
• Time Series Analysis: Smoothing, Averaging, Autoregression, Theta
• Data Analysis/Preparation: Hypothesis testing, correlation analysis, data visualization, dimensionality reduction, class balancing, imputation, text processing