Below is a collection of my data analysis and machine learning projects. Each one demonstrates my ability to translate raw data into actionable insights and solutions.
Analyze the Superstore dataset to find key business insights and translate them into actionable recommendations, backed by data analysis and supported by an interactive dashboard.
Skills:
Findings: Uncovered that the Tables product line was operating at a -8.4% profit margin, costing the company over $64,000 annually. The root cause was identified as excessive discounting (~50%) in specific markets (EU South, APAC Southeast Asia).
Analyze transaction data to identify the most effective machine learning model for accurately detecting fraudulent transactions while minimizing false positives.
Skills:
Findings: After comparing various models, the SVM model was the most effective, correctly identifying 97% of fraudulent cases while minimizing false positives. This is critical for fraud detection, as it prevents unnecessary customer transaction declines while stopping fraudulent activity.
Compare convolutional neural networks (CNN) and support vector machines (SVM) for accurate traffic light classification to ensure safety in autonomous driving systems.
Skills:
Findings: Model evaluation against strict safety criteria showed the SVM model was the optimal solution. It achieved zero misclassifications for red lights, the requirement, while maintaining 99%+ accuracy on all other signals.
Explored Boulder public crime data to identify temporal patterns, verify public reports of rising crime, and uncover seasonal trends to inform better public safety strategies.
Skills:
Findings: Uncovered a consistent seasonal peak in crime during August, strongly correlated with higher temperatures and the annual influx of students, providing data-driven insights for proactive public safety planning.
A comprehensive redesign of a public safety dashboard to improve visual cohesion and storytelling through spatial analysis and consistent visual design
Skills:
Findings: Redesigned dashboard revealed a 200% increase in shootings (2017-2021) and clear spatial correlations between neighborhood demographics and shooting incidents.
My Role: Research Analyst & Framework Specialist
A collaborative analysis of hurricane evacuation research where I applied a micro-meso-macro framework to synthesize insights from key literature and contribute to group findings.
My Contributions:
Skills:
Findings: Disaster response failures are driven by lost insurance documentation during evacuation, inaccurate storm predictions causing misaligned evacuation plans, the lack of a unified national recovery system delaying federal aid, and power outages that disrupt rescue communications.
My Role: Primary Researcher & Systems Analyst
Original research investigating technology infrastructure and operational workflows through stakeholder interviews and systems analysis.
My Contributions:
Skills:
Findings: There are critical software integration challenges and communication gaps affecting operational efficiency in animal welfare services.