Statistical Ideas in Code
One cleaned-up page per classic paper, with math and minimal Python demos
This site is a compact companion to the STAT 319 journal-club collection. Each page distills the paper’s core contribution, writes the central mathematical object, implements a minimal demo using NumPy/SciPy/SymPy/Matplotlib, and links to common production implementations.
The demos are intentionally small. They are meant to make the mechanism visible, not to replace mature libraries.
| Paper | Core idea | Page |
|---|---|---|
| Laird (1982), Random-Effects Models for Longitudinal Data | Model repeated outcomes with subject-level random variation. | Random effects |
| Benjamini and Hochberg (1995), Controlling the False Discovery Rate | Control the expected share of false discoveries among discoveries. | FDR |
| Breiman (2001), Statistical Modeling: The Two Cultures | Contrast explanatory stochastic models with predictive algorithmic models. | Two cultures |
| Efron (1979), Bootstrap Methods: Another Look at the Jackknife | Approximate sampling distributions by resampling the empirical distribution. | Bootstrap |
| Dempster, Laird, and Rubin (1977), Maximum Likelihood from Incomplete Data via the EM Algorithm | Optimize likelihoods with latent or missing data by alternating E and M steps. | EM |
| Baum and Welch (1970), A Maximization Technique… | Fit hidden Markov models with forward-backward EM. | Baum-Welch |
| Gelfand and Smith (1990), Sampling-Based Approaches to Calculating Marginal Densities | Estimate posterior marginals by simulation rather than closed-form integration. | MCMC |
| Cox (1972), Regression Models and Life-Tables | Estimate hazard ratios with a partial likelihood that leaves the baseline hazard unspecified. | Cox model |
| Rosenbaum and Rubin (1983), The Central Role of the Propensity Score | Reduce treatment-assignment confounding to a scalar balancing score. | Propensity scores |
| Hastie and Tibshirani (1986), Generalized Additive Models | Replace a linear predictor with a sum of smooth functions. | GAM |
| Shafer and Vovk (2008), A Tutorial on Conformal Prediction | Wrap predictions with finite-sample-valid uncertainty sets under exchangeability. | Conformal prediction |
| Friedman (1999), Greedy Function Approximation | Build additive models by functional gradient descent. | Gradient boosting |
| Robins (1999), Association, Causation, and Marginal Structural Models | Use inverse-probability weighting for time-varying treatments and confounders. | MSMs |
| Tibshirani (1996), Regression Shrinkage and Selection via the Lasso | Use an \(\ell_1\) penalty for simultaneous shrinkage and variable selection. | Lasso |