Zach Wilde

Data Scientist | M.S. Statistics

I build predictive models and extract insight from complex data.

Resume → | LinkedIn →

Featured Work

Predictive Modeling of Social Media Engagement

Master’s Thesis · 7,000 TikTok Videos · 40+ Engineered Features

Built an end-to-end modeling pipeline combining metadata, OCR, audio transcription, and temporal feature engineering to identify drivers of content exposure. Compared OLS, Ridge, and Lasso under repeated train/test validation.

What I learned: Feature engineering mattered more than model choice.

Read Case Study →

Classifying Crime in US Cities

PCA · LDA · QDA · Classification

Used dimensionality reduction and discriminant analysis to classify communities as high versus low crime based on socioeconomic indicators.

Result: ~81% out-of-sample accuracy.

Read Case Study →

Predicting Religion from Country Flag

Multinomial Regression · Decision Trees

Compared multiple classification approaches to predict dominant religion using flag characteristics and geographic information.

Result: Explored interpretability versus predictive performance across model families.

Read Case Study →

What I Work On

• Statistical Modeling
• Predictive Analytics
• R & Python
• Experimental Design
• Pricing & Revenue Analytics
• Healthcare Data
• Feature Engineering
• Data Pipelines

About

I’m a statistician and data scientist based in Salt Lake City. My background spans healthcare analytics, pricing modeling, media, telecommunications, software data migration, and graduate-level statistical research.

I recently completed my M.S. in Statistics at the University of Utah, where my thesis focused on modeling engagement in short-form social media using multimodal feature engineering and regularized regression.

When I’m not working with data, I’m usually building something - whether that’s a Raspberry Pi NAS, a 3D printed hardware project, or a new analytics workflow.