Super Data Science: ML & AI Podcast with Jon Krohn 793: Bayesian Methods and Applications, with Alexandre Andorra
Jun 18, 2024
Alexandre Andorra, Co-founder of PyMC Labs, joins Jon Krohn to discuss Bayesian methods, PyMC, PyStan, NumPyro, ArviZ, and Gaussian Processes. They cover practical Bayesian statistics, epistemology, consulting services, post-modeling diagnostics, and visualization. Learn about Bayesian modeling, hierarchical modeling, and Gaussian Processes in data analysis, as well as book recommendations and stoic philosophy.
AI Snips
Chapters
Transcript
Episode notes
Bayesian Power
- Bayesian statistics excels with limited, unreliable data and strong prior knowledge.
- It's powerful for various fields, from marketing to sports modeling.
Electoral Forecasting
- Alex Andorra's interest in Bayesian stats started with electoral forecasting, driven by noisy polls.
- This led him to incorporate domain knowledge (prior information) for better predictions.
Explicit Priors
- Bayesian priors force assumptions to be explicit, which facilitates better discussions.
- This is crucial for models explaining social or political phenomena, where biases are common.