The Duct Tape of Heroes: Bayes Rule
summary
Shows concrete claims and demos: uses pomegranate probabilistic programming, builds Bayesian probabilistic graphs for disease testing, model selection, feature generation, operations research and a Heroes of the Storm dataset, and asserts Bayesian models outperform random forests on that game data. Presents code examples (pandas sampling, beta posteriors, KDE for guided search), handles missing data at learn and predict time, and measures how posterior beliefs converge as sample size increases. Slides and dataset links provided.
AI-written summary · about
chapters (8)
- 0:00Bayesian thinking
- 1:05Bayes rule explained
- 7:22Graphical models in Python
- 15:49Biased data in Bayes
- 22:04Bayesian optimization
- 26:58Heroes match predictions
- 34:50Bayes vs frequentist vs ML
- 38:00Closing remarks
related talks
42mThe Duct Tape of Heroes Bayesian statistics
01h 03mAn Overview of Probabilistic Programming
43mProbabilistic Programming in the Real World
29mHow to use Data Science Superpowers in real life, a Bayesian perspective
01h 27mProbabilistic Python: An Introduction to Bayesian Modeling with PyMC
42mBayes and Bool Walk into a Bar
43mThe Power of Probabilistic Programming
42mBayesDB: Query the Probable Implications of Data