Data 145: Evidence and Uncertainty - Topic Map

Created on April 23, 2026

2026   ·   statistics   data145   notes   statistics-inference   ·   technical-blogs

My notes on Data 145: Evidence and Uncertainty. This page is a compact topic map for future review and for anyone else looking for a high-level reference.

Course thread: how to turn noisy data into evidence, quantify uncertainty, and make decisions without pretending randomness disappeared.

Phase 1 Notes

Phase 1 notes Data 145 Phase 1: From MLE to Neyman-Pearson to Reward Models A long-form roadmap for estimation, Fisher information, bootstrap, hypothesis testing, Neyman-Pearson, and connections to modern AI.

Phase 2 Notes

Phase 2 notes Lectures 14-15: Testing, Confidence, and Power Testing structure, confidence intervals, and power. Phase 2 notes Lectures 17-18: Likelihood and Decision Rules Likelihood reasoning and general testing procedures. Phase 2 notes Lectures 19-20: Multiple Testing Bonferroni, FWER, FDP, and FDR. Phase 2 notes Lectures 21-22: MGFs and Concentration MGFs, Chernoff thinking, Hoeffding, and tail bounds. Phase 2 notes Lectures 23-24: Geometry of Linear Models Rotations, nuisance/signal/residual blocks, t, chi-squared, F, ANOVA, and regression. Phase 2 notes Lectures 25-26: Gibbs Sampling and Hierarchical Bayes MCMC, Gibbs full conditionals, burn-in, shrinkage, empirical Bayes, and hyperparameter concentration. Phase 2 notes Lecture 27: Introduction to Causal Inference Potential outcomes, randomized trials, ATE estimation, confounding, propensity scores, and inverse propensity weighting.
Study path: use Phase 1 for the pre-midterm course narrative, then use Phase 2 as post-midterm topic-by-topic references.


Enjoy Reading This Article?

Here are some more articles you might like to read next:

  • From Memory to Photonics: Solving the Next Bottleneck in AI Scaling
  • Training a Language Model from Scratch (Part 2: FlashAttention and Device Memory)
  • Training a Language Model from Scratch (Part 1: Building Blocks)
  • Data 145 Phase 1: From MLE to Neyman-Pearson to Reward Models
  • Explainable AI (XAI) and Model Interpretability (SHAP, Integrated Gradients, and Sparse Autoencoders)