Probabilistic-Programming on Ryan Orban

Stochastic Volatility Modeling with PyMC

me@ryanorban.com (Ryan Orban) — Wed, 08 Jan 2014 00:00:00 +0000

Summary

This PyMC example notebook demonstrates stochastic volatility modeling — a Bayesian inference approach to modeling time-varying variance in financial time series. Standard financial models (like Black-Scholes) assume constant volatility, but real markets have volatility clustering: calm periods and turbulent periods that persist in clusters. Stochastic volatility models treat volatility itself as a latent random process that evolves over time.

BayesDB

me@ryanorban.com (Ryan Orban) — Tue, 10 Dec 2013 00:00:00 +0000

Summary

BayesDB is a probabilistic database system developed at MIT CSAIL’s Probabilistic Computing Project (probcomp). The core premise: analysts should be able to ask statistical questions about data using a familiar SQL-like interface without needing to choose, fit, and diagnose a statistical model manually. BayesDB handles the model selection internally.

Bayesian Methods for Hackers

me@ryanorban.com (Ryan Orban) — Thu, 06 Jun 2013 00:00:00 +0000

Summary

Cameron Davidson-Pilon’s Probabilistic Programming and Bayesian Methods for Hackers is an open-source Jupyter notebook-based book that teaches Bayesian inference computationally rather than mathematically. The central premise: most programmers can intuit Bayesian reasoning (update beliefs given evidence) but get stuck on the math. By starting with PyMC3 (a Python probabilistic programming library) and building intuition from code and visualizations, the book sidesteps the derivation-heavy approach of classical Bayesian textbooks.