Trying to clean up my notes here — definitely I haven’t been doing as much technical work as before, but things may always come in handy. Just putting these things out here. This is a random mix-bag of statical methods that I have read or used in the last few years:
- A Gentle Introduction to Maximum Likelihood Estimation
- Multivariate Gaussian Distribution
- NLP in R
- A Literature Review on Uplift Modeling
- Counterfactual Estimation and Optimization of Click Metrics
- Hierarchical Liner Regression in PyMC3 (Bayesian) and Hierarchical Partial Pooling, power of Bayesian A/B Testing, Chris Stucchio’s long paper, and intuitive explanations, bayesAB (R), conversion rates with Bayes Rule, another blog, general rstanarm intro
- CUPED — covariance adjustment
- Alpha Spending Function on Sequential Analysis [ftp://maia-2.biostat.wisc.edu/pub/chappell/641/papers/paper35.pdf https://eclass.uoa.gr/modules/document/file.php/MATH301/PracticalSession3/LanDeMets.pdf] and interim analysis, always valid-p-value, peeking AB tests, Booking.com and multi-variate tests, platform article on online controlled experiments, always valid inference
- Delta method — approximate Mean and Variance of a Ratio
- Interesting learning Budget within session of recommender systems
- Clustering: clusterR package
- Generalized Linear Models
- Working-Hotelling procedure (Simultaneous Inference)
- Plotting interaction terms of coefficients, effects package
- Propensity Score Matching , R Tutorial,
- Deep Reinforcement Learning Bootcamp (2017)
- Modeling Covariance Structures
- A framework for Multi-Armed Bandit Testing
- Advanced R - Ebook
- Death of the Statistical Test of Hypothesis
- Seemingly Unrelated Regressions
- Double / Debiased ML
- Imperfect Treatment Assignment in Online Experiments
- Kaplan-Meier Estimator
- STAN modeler
- Mantel-Haenszel
- Confidence vs Credibility Intervals
- Dunnett’s test: simultaneous confidence intervals of 1 v.s. many tests
- Analysis of Multivariate time series using the MARSS package
- Density distribution R package
- Flexible imputation methods
- EM algorithm for Mixtures of Multinomials
- Fitting data to distributions, estimation of Dirchelt-Multinomial Distributions
- Expectation-Maximization Multinomial, MLE for Incomplete Data
- JSM2019 Bayesian workshop materials (R package brms)
- Limiting Bias from test-control interference in Online Marketplace Experiments
- Additional interesting R packages (ggeconodist, brms, change point hypothesis testing), DAG simulation, plotting Bayes Factors, assess regression model perf, dagitty, ggdag, pdqr, correlationfunnel, skimr, visDat, naniar, UpSetR, mlr
- Causal Effects
- Meta Regression: Network Meta-analysis, nmaINLA, Meta-Regression
- Empirical Bayes
- Athey’s Matrix Completion Methods
- Two sample quantile comparison
- R-packages Book, and p-value functions