Ben Barnard

Data Scientist

Wells Fargo


Ben Barnard is a Data Scientist for Wells Fargo and a Statistics Instructor for the University of South Alabama. His research interests include Bayesian methods, statistical programming, teaching statistics, and using common sense in data science. He currently leads the development of several R and python packages for internal Wells Fargo use. Outside of work Ben likes to spend time with his family, working on the house, competing in triathlons, and speaking at conferences.


  • Bayesian Methods
  • Statistical Programming
  • Data Science Consulting


  • PhD in Statistical Sciences, 2018

    Baylor University

  • MS in Statistical Sciences, 2016

    Baylor University

  • MSEd in Exercise Physiology, 2013

    Baylor University

  • BS in Exercise Science, 2011

    University of Texas at Arlington



Data Scientist

Wells Fargo

May 2018 – Present Remote
Responsibilities include:

  • Model development lead for Human Capital Key Risk Indicators using time series models.
  • Developer of 20 internal R packages using agile development with 2-week sprints and daily asynchronous stand-ups in Skype chat groups.
  • Enterprise Product Steward for R and Rtools monitoring intake and maintenance of new versions, and mitigating security vulnerabilities for ~2500 installs across the enterprise.
  • Lead transformation of team to use software development life-cycle appropriate tools such as GitHub, Jenkins, and Artifactory.
  • Model developer for turnover model using random forest to inform a Bayesian joint estimable logistic regression and cox proportional hazard model in stan.

Statistics Instructor

University of South Alabama

Aug 2017 – Present Mobile, Alabama
Responsibilities include:

  • Coordinator for design and instruction of Applied Statistics for Health Sciences courses.
  • Quality matters and Team Based Learning instruction certifications.
  • Developer of tumor image classification model using a convolution neural network in tensorflow.

Biostatistics Intern

Baylor Scott & White Health

Mar 2017 – Aug 2017 Temple, Texas
Responsibilities include:

  • Model developer of physician quality incentive programs random forest model.
  • Statistical analysis of Necrotizing Enterocolitis retrospective study using logistic regression mixed model in PROC GLIMMIX.
  • Bayesian sample size simulation study in stan for Family, Food and Fun program.

Graduate Assistant

Baylor University

May 2013 – Aug 2017 Waco, Texas
Responsibilities include:

  • Developer for conversion rate key performance indicators on graduate applications.
  • Model developer for student success gradient boosting machine model using XGBoost.
  • Designer of A/B testing for the Graduate School website and application pages.
  • Model developer for applicant and graduate student segmentation clustering models using k-means clustering.
  • Developer for automated reporting on applications and enrollment data using D3.js, SAS reporting studio, Shiny server.
  • Model developer for athlete injury prediction model using recurrent neural network in tensorflow.
  • Statistical anaylsis on paleosol composition using semi-parametric models in PROC TRANSREG.

Graduate Teaching Assistant

Baylor University

Aug 2011 – May 2013 Waco, Texas
Responsibilities include:

  • Instruction and preparation of Human, Health, Performance, and Recreation department activity and health courses.
  • Experimental design and analysis of dehydration training studies using analysis of variance and semiparmetric regression.
  • Development of periodized training plans for clinical populations using frequency, intensity, time, and type principles.

Recent Posts

R Packages

covTestR: Covariance Matrix Tests

Testing functions for Covariance Matrices. These tests include high-dimension homogeneity of covariance matrix testing described by …

likelihoodExplore: Likelihood Exploration

Provides likelihood functions as defined by Fisher (1922) doi:10.1098/rsta.1922.0009 and a function that creates likelihood functions …

rWishart: Random Wishart Matrix Generation

An expansion of R’s ‘stats’ random wishart matrix generation. This package allows the user to generate singular, …

Recent & Upcoming Talks