Bayesian thinking where uncertainty meets decisions.
Learning Bayesian Statistics is a technical interview show about Bayesian thinking in practice. Episodes move between modeling details and real-world judgment, from risk-aware pricing and synthetic control to probabilistic programming, adaptive trials, and communicating uncertainty to stakeholders.
PRESS PLAY
Learning Bayesian Statistics is a technical conversation series about Bayesian inference as it is used, argued over, and explained in practice. The show has a strong modeling center. It returns often to posteriors, priors, utility, uncertainty, causal effects, computation, and the mental shift from single-number prediction to belief updating. It also keeps asking what happens after the model is built. That practical edge matters. Episodes on pricing under uncertainty separate the modeling problem from the decision problem, showing how risk aversion can change an apparently optimal recommendation. Conversations on probabilistic programming connect research ideas such as inference compilation, amortized inference, surrogate networks, and Python tooling to the stubborn realities of enterprise adoption. Sometimes the barrier is not the math. Sometimes it is the spreadsheet. The causal inference episodes are especially focused on how probabilistic framing helps stakeholders reason through nuanced business questions. Synthetic control, synthetic difference-in-differences, credible intervals, posterior distributions, and false-positive risk become tools for making better decisions rather than ritual outputs. Clinical trial discussions bring the same philosophy into biostatistics, where adaptive designs, multiple arms, dose changes, and response-adaptive randomization can make old analytical habits feel too rigid. The show also makes room for computation at scale, including INLA, MCMC comparisons, spatial statistics, survival analysis, penalized complexity priors, and future work around faster solvers and GPU acceleration. Not every episode stays inside statistics departments. Some conversations move into tech careers, AI, return-to-office debates, leadership, resilience, and the pressure on professionals to understand how their companies make money. The through-line is uncertainty. Learning Bayesian Statistics is best for listeners who want technical substance without losing sight of decisions, incentives, and communication. It assumes curiosity and rewards patience. It is not a quick tips show. It is a place where Bayesian methods are treated as living tools for science, business, medicine, and work.
Made for: Designed for researchers, data scientists, analysts, statisticians, and technically curious professionals who want Bayesian methods tied to real modeling work. It also suits decision makers who need clearer language for uncertainty.
What sets it apart: The show pairs detailed statistical discussion with applied decision pressure: pricing, causal inference, clinical trials, enterprise adoption, and stakeholder communication. It makes Bayesian inference feel less like a doctrine and more like a disciplined way to act under uncertainty.
Are you a researcher or data scientist / analyst / ninja? Do you want to learn Bayesian inference, stay up to date or simply want to understand what Bayesian inference is? Then this podcast is for you! You'll hear from researchers and practitioners of all fields about how they use Bayesian statistics, and how in turn YOU can apply these methods in your modeling workflow. When I started learning Bayesian methods, I really wished there were a podcast out there that could introduce me to the methods, the projects and the people who make all that possible. So I created "Learning Bayesian Statistics", where you'll get to hear how Bayesian statistics are used to detect black matter in outer space,…
Excel becomes Bayesian territory, while Alex Andorra helps you translate uncertainty for the CFO.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A practical look at bringing probabilistic programming into enterprise workflows through Excel.
Excel gives this single-episode review an unusually practical entry point into Bayesian inference, with Andreas Munch positioned between probabilistic programming research, amortized inference, and enterprise workflows.
Big ideas, louder than necessary.
Behavior analysis, poured into real practice.
One zombie metaphor, stretched from parasites to school.
This episode discusses Bayesian decision theory for pricing under uncertainty, contrasting risk-neutral and risk-averse approaches.
Learning Bayesian Statistics episode 155, recorded March 19, 2026, centers on probabilistic programming, amortized inference, and the claim that the barrier to Bayesian adoption can be the file format.
A data scientist explains how Bayesian methods provide nuanced uncertainty quantification for causal inference in business contexts, enabling better decision-making.
Learning Bayesian Statistics #154 features Thomas Pinder, whose work spans Gaussian processes, experimentation, and causal inference at Amazon, Uber, and Netflix.
Exploring the neuroscience behind donor decision making and how behavioral science and ethical AI can improve philanthropy.
A guest explains how mass layoffs, AI competition, and return-to-office policies changed psychological safety in Silicon Valley workplaces.
Alana Karen discusses the technology industry after 23 years at Google and the publication of The Hard Tech Era.
Learning Bayesian Statistics features Scott Berry discussing how Berry Consultants moved Bayesian methods into clinical-trial design after its founding with Donald Berry in 2000.
Scott Berry shows how three Alzheimer's drug trials sharing one placebo arm eliminate 1,000 placebo patients without losing statistical power.
Martin Ingram says ADVI users end up disillusioned because the supposedly faster method means you now "have to babysit it."
An interview with Haavard Rue and Janet van Niekerk about the INLA method for efficient Bayesian inference in large-scale latent Gaussian models.
Teemu Säilynoja explains simulation-based calibration (SBC) and its importance in Bayesian patient modeling.
Teemu Säilynoja discusses simulation-based calibration and visual predictive checking for validating Bayesian models.
Sean Pinkney and Adrian Seyboldt discuss the zero-subnormal project in Stan, which improves the efficiency and flexibility of hierarchical Bayesian models.
Tom Griffiths discusses how human intelligence is shaped by constraints like limited data and fixed compute, contrasting it with AI systems.