Statistical Consulting & Research

Joseph S. Miller

Designing efficient experiments and statistical modeling strategies.

About

I'm a statistician developing efficient optimal experiment design tools that work for survey sampling, A/B testing, and other data collection challenges. My work blends new frameworks in information theory with applied modeling to quantify uncertainty and make adaptive experiment design practical. I also specialize in tractable probabilistic modeling, enabling efficient probabilistic queries and inference.

Consulting

Typical collaborations include:

Efficient experiment design Survey sampling strategy A/B testing pipelines Bayesian modeling Tractable probabilistic modeling

Papers

Similarity-Sensitive Entropy: Induced Kernels and Data-Processing Inequalities (arXiv:2601.03064).

Notes

Technical notes and case studies, including:

Kernalized coarse graining.

Mutual information for Bayesian experimental design.