Categorical regression / Fitting independent trees |
Categorical regression / Model Specification |
Modeling Heteroscedasticity with BART / Model Specification |
Bayesian Additive Regression Trees: Introduction / Biking with BART |
Bayesian Additive Regression Trees: Introduction / Coal mining with BART |
Bayesian Additive Regression Trees: Introduction / Biking with BART / Out-of-Sample Predictions / Regression |
Bayesian Additive Regression Trees: Introduction / Biking with BART / Out-of-Sample Predictions / Time Series |
Quantile Regression with BART / Asymmetric Laplace distribution |
Bayesian Estimation Supersedes the T-Test / Example: Drug trial evaluation |
Confirmatory Factor Analysis and Structural Equation Models in Psychometrics / Full Measurement Model |
Confirmatory Factor Analysis and Structural Equation Models in Psychometrics / Measurement Models / Intermediate Cross-Loading Model |
Confirmatory Factor Analysis and Structural Equation Models in Psychometrics / Measurement Models |
Confirmatory Factor Analysis and Structural Equation Models in Psychometrics / Bayesian Structural Equation Models / Model Complexity and Bayesian Sensitivity Analysis |
Generalized Extreme Value Distribution / Inference |
Estimating parameters of a distribution from awkwardly binned data / Example 2: Parameter estimation with the other set of bins / Model specification |
Estimating parameters of a distribution from awkwardly binned data / Example 6: A non-normal distribution / Model specification |
Estimating parameters of a distribution from awkwardly binned data / Example 3: Parameter estimation with two bins together / Model Specification |
Estimating parameters of a distribution from awkwardly binned data / Example 4: Parameter estimation with continuous and binned measures / Model Specification |
Estimating parameters of a distribution from awkwardly binned data / Example 5: Hierarchical estimation / Model specification |
Estimating parameters of a distribution from awkwardly binned data / Example 1: Gaussian parameter estimation with one set of bins / Model specification |
Factor analysis / Model / Alternative parametrization |
Factor analysis / Model / Direct implementation |
Hierarchical Partial Pooling / Approach |
NBA Foul Analysis with Item Response Theory / Sampling and convergence |
Probabilistic Matrix Factorization for Making Personalized Recommendations / Probabilistic Matrix Factorization |
Model building and expansion for golf putting / A new model |
Model building and expansion for golf putting / Fitting the distance angle model |
Model building and expansion for golf putting / Fitting the model on the new data |
Model building and expansion for golf putting / Logit model |
Model building and expansion for golf putting / Geometry-based model / Prior Predictive Checks |
Fitting a Reinforcement Learning Model to Behavioral Data with PyMC / Estimating the learning parameters via PyMC / Alternative model using Bernoulli for the likelihood |
Fitting a Reinforcement Learning Model to Behavioral Data with PyMC / Estimating the learning parameters via PyMC |
Reliability Statistics and Predictive Calibration / Bayesian Modelling of Reliability Data / Direct PYMC implementation of Weibull Survival |
A Hierarchical model for Rugby prediction / Building of the model |
Simpson’s paradox / Model 1: Pooled regression / Conduct inference |
Simpson’s paradox / Model 2: Unpooled regression with counfounder included / Conduct inference |
Simpson’s paradox / Model 3: Partial pooling model with confounder included / Conduct inference |
Introduction to Bayesian A/B Testing / Bernoulli Conversions / Data |
Introduction to Bayesian A/B Testing / Generalising to multi-variant tests |
Introduction to Bayesian A/B Testing / Value Conversions |
Bayesian Non-parametric Causal Inference / Causal Inference and Propensity Scores / Double/Debiased Machine Learning and Frisch-Waugh-Lovell / Applying Debiased ML Methods |
Bayesian Non-parametric Causal Inference / Causal Inference and Propensity Scores / Mediation Effects and Causal Structure |
Bayesian Non-parametric Causal Inference / Causal Inference and Propensity Scores / Non-Confounded Inference: NHEFS Data / Propensity Score Modelling |
Bayesian Non-parametric Causal Inference / Causal Inference and Propensity Scores / Non-Confounded Inference: NHEFS Data / Regression with Propensity Scores |
Difference in differences / Bayesian difference in differences / Inference |
Counterfactual inference: calculating excess deaths due to COVID-19 / Inference |
Interrupted time series analysis / Inference |
Bayesian mediation analysis / Define the PyMC model and conduct inference |
Bayesian mediation analysis / Double check with total effect only model |
Does the effect of training upon muscularity decrease with age? / Define the PyMC model and conduct inference |
Regression discontinuity design analysis / Inference |
Bayes Factors and Marginal Likelihood / Savage-Dickey Density Ratio |
Model Averaging / Weighted posterior predictive samples |
Sampler Statistics / Multiple samplers |
Sampler Statistics |
Using Data Containers / Applied example: height of toddlers as a function of age |
Using Data Containers / Applied Example: Using Data containers as input to a binomial GLM |
Using Data Containers / Using Data Containers for readability and reproducibility / Named dimensions with data containers |
Using Data Containers / Using Data Containers to mutate data / Using Data container variables to fit the same model to several datasets |
Using Data Containers / Using Data Containers for readability and reproducibility |
Baby Births Modelling with HSGPs / EDA and Feature Engineering / Model Fitting and Diagnostics |
Kronecker Structured Covariances / LatentKron / Model |
Gaussian Processes: Latent Variable Implementation / Example 1: Regression with Student-T distributed noise / Coding the model in PyMC |
Gaussian Processes: Latent Variable Implementation / Example 2: Classification |
Marginal Likelihood Implementation / Example: Regression with white, Gaussian noise |
Student-t Process / Poisson data generated by a T process |
Example 1: A hierarchical HSGP, a more custom model / Example 2: An HSGP that exploits Kronecker structure / Sampling & Convergence checks |
Example 1: A hierarchical HSGP, a more custom model / Looking for a beginner’s introduction? / Sampling & Convergence checks |
Gaussian Processes: HSGP Reference & First Steps / Example 1: Basic HSGP Usage / Define and fit the HSGP model |
Gaussian Processes: HSGP Reference & First Steps / Example 1: Basic HSGP Usage / Example 2: Working with HSGPs as a parametric, linear model / Results |
Inference |
Binomial regression / Binomial regression model |
Discrete Choice and Random Utility Models / Choosing Crackers over Repeated Choices: Mixed Logit Model |
Discrete Choice and Random Utility Models / Experimental Model: Adding Correlation Structure |
Discrete Choice and Random Utility Models / Improved Model: Adding Alternative Specific Intercepts |
Discrete Choice and Random Utility Models / The Basic Model |
Hierarchical Binomial Model: Rat Tumor Example / Computing the Posterior using PyMC |
2. ModelA: Auto-impute Missing Values / 2.3 Sample Posterior, View Diagnostics / 2.3.1 Sample Posterior and PPC |
1. Model0: Baseline without Missing Values / 1.3 Sample Posterior, View Diagnostics / 1.3.1 Sample Posterior and PPC |
GLM: Model Selection / Generate toy datasets / Demonstrate simple linear model / Define model using explicit PyMC method |
GLM: Negative Binomial Regression / Negative Binomial Regression / Create GLM Model |
2. Model B: A Better Way - Dirichlet Hyperprior Allocator / 2.3 Sample Posterior, View Diagnostics / 2.3.1 Sample Posterior and PPC |
1. Model A: The Wrong Way - Simple Linear Coefficients / 1.3 Sample Posterior, View Diagnostics / 1.3.1 Sample Posterior and PPC |
Ordinal Scales and Survey Data / Fit a variety of Model Specifications / Bayesian Particularities |
Ordinal Scales and Survey Data / Liddell and Kruschke’s IMDB movie Ratings Data |
Out-Of-Sample Predictions / Define and Fit the Model |
GLM: Poisson Regression / Poisson Regression / 1. Manual method, create design matrices and manually specify model |
Setup / 5. Linear Model with Custom Likelihood to Distinguish Outliers: Hogg Method / 5.2 Fit Model / 5.2.1 Sample Posterior |
Setup / 4. Simple Linear Model with Robust Student-T Likelihood / 4.2 Fit Model / 4.2.1 Sample Posterior |
Setup / 3. Simple Linear Model with no Outlier Correction / 3.2 Fit Model / 3.2.1 Sample Posterior |
GLM: Robust Linear Regression / Robust Regression / Normal Likelihood |
Rolling Regression / Rolling regression |
Rolling Regression |
Bayesian regression with truncated or censored data / Run the truncated and censored regressions |
Bayesian regression with truncated or censored data / The problem that truncated or censored regression solves |
A Primer on Bayesian Methods for Multilevel Modeling / Adding group-level predictors |
A Primer on Bayesian Methods for Multilevel Modeling / Conventional approaches |
A Primer on Bayesian Methods for Multilevel Modeling / Adding group-level predictors / Correlations among levels |
A Primer on Bayesian Methods for Multilevel Modeling / Non-centered Parameterization |
A Primer on Bayesian Methods for Multilevel Modeling / Partial pooling model |
A Primer on Bayesian Methods for Multilevel Modeling / Varying intercept and slope model |
A Primer on Bayesian Methods for Multilevel Modeling / Varying intercept model |
LKJ Cholesky Covariance Priors for Multivariate Normal Models |
Bayesian Missing Data Imputation / Bayesian Imputation |
Bayesian Missing Data Imputation / Bayesian Imputation by Chained Equations / PyMC Imputation |
Using a “black box” likelihood function / Comparison to equivalent PyMC distributions |
Using a “black box” likelihood function / PyTensor Op with gradients / Model definition |
Using a “black box” likelihood function / Introduction |
Using a “black box” likelihood function / PyTensor Op without gradients / Model definition |
Using a “black box” likelihood function / Using a Potential instead of CustomDist |
Bayesian copula estimation: Describing correlated joint distributions / PyMC models for copula and marginal estimation |
How to debug a model / Introduction / Bringing it all together |
How to debug a model / Introduction / Troubleshooting a toy PyMC model |
Automatic marginalization of discrete variables / Coal mining model |
Automatic marginalization of discrete variables / Gaussian Mixture model |
Using ModelBuilder class for deploying PyMC models / Standard syntax |
Splines / The model / Fit the model |
Splines / Predicting on new data |
Updating Priors / Words of Caution / Model specification |
How to wrap a JAX function for use in PyMC / Wrapping the JAX function in PyTensor / Sampling with PyMC |
General API quickstart / 3. Inference / 3.2 Analyze sampling results |
General API quickstart / 4. Posterior Predictive Sampling |
General API quickstart / 4.1 Predicting on hold-out data |
General API quickstart / 3. Inference / 3.1 Sampling |
Dirichlet mixtures of multinomials / Dirichlet-Multinomial Model - Explicit Mixture |
Dirichlet mixtures of multinomials / Dirichlet-Multinomial Model - Marginalized |
Dirichlet mixtures of multinomials / Multinomial model |
Dirichlet process mixtures for density estimation / Dirichlet process mixtures |
Gaussian Mixture Model |
ODE Lotka-Volterra With Bayesian Inference in Multiple Ways / Gradient-Free Sampler Options / DE MetropolisZ Sampler |
ODE Lotka-Volterra With Bayesian Inference in Multiple Ways / Gradient-Free Sampler Options / DEMetropolis Sampler |
ODE Lotka-Volterra With Bayesian Inference in Multiple Ways / Bayesian Inference with Gradients / Simulate with Pytensor Scan / Inference Using NUTs |
ODE Lotka-Volterra With Bayesian Inference in Multiple Ways / Bayesian Inference with Gradients / PyMC ODE Module / Inference with NUTS |
ODE Lotka-Volterra With Bayesian Inference in Multiple Ways / Gradient-Free Sampler Options / Metropolis Sampler |
ODE Lotka-Volterra With Bayesian Inference in Multiple Ways / Gradient-Free Sampler Options / Slice Sampler |
DEMetropolis and DEMetropolis(Z) Algorithm Comparisons / Helper Functions / Sampling |
DEMetropolis(Z) Sampler Tuning / Conclusions |
DEMetropolis(Z) Sampler Tuning / Helper Functions / Sampling |
Lasso regression with block updating |
Compound Steps in Sampling / Compound steps |
Compound Steps in Sampling / Compound steps by default |
Compound Steps in Sampling / Order of step methods |
Compound Steps in Sampling / Specify compound steps |
Using a custom step method for sampling from locally conjugate posterior distributions / Comparing partial conjugate with full NUTS sampling |
Conditional Autoregressive (CAR) model / Writing some models in PyMC / Our first model: an independent random effects model |
Conditional Autoregressive (CAR) model / Writing some models in PyMC / Our second model: a spatial random effects model (with fixed spatial dependence) |
Conditional Autoregressive (CAR) model / Writing some models in PyMC / Our third model: a spatial random effects model, with unknown spatial dependence |
Different Covariance Functions |
Model Specification |
Demonstrating the BYM model on the New York City pedestrian accidents dataset / Sampling the model |
Linear Regression / (5) Analyse real data |
Linear Regression / Simulation-based Validation & Calibration |
Categories / Analyze real sample / Analyze the synthetic people |
Curves from lines / Example: Cherry Blossom Blooms / Draw some samples from the prior |
Categories / Testing / Fit total effect on the synthetic sample |
Curves from lines / Polynomial Linear Models / Fitting N-th Order Polynomials to Height / Width Data |
BONUS: Full Luxury Bayes / Why would we do this? |
The Periodic Table of Confounds / Fork Example: Marriage Rates & Divorce Rates (& Waffle House!) / (3) Statistical Model for Causal Effect of Marriage on Divorce / Run the statistical model on the simulated data |
The Periodic Table of Confounds / Continuous Example / Statistical Model for Causal Effect of Age on Divorce Rate |
Confounds / Backdoor Criterion / Unstratified (confounded) Model / Fit the unstratified model, ignoring Z (and U) |
Confounds / Backdoor Criterion / Stratifying by Z (unconfounded) |
Infinite causes, finite data / Penalty Prediction & Model (Mis-) Selection / Simulate the plant growth experiment / Correct adjustment set (not stratifying by F) |
Infinite causes, finite data / Outliers and Robust Regression / Fit Least Square Model |
Infinite causes, finite data / Penalty Prediction & Model (Mis-) Selection / Simulate the plant growth experiment / Incorrect adjustment set (stratifying by F) |
Infinite causes, finite data / Robust Linear Regression Using the Student-t Likelihood |
Drawing the Markov Owl 🦉 / Including Judge Effects / Fit the judge model |
2012 New Jersey Wine Judgement / Simplest Model / Fit the simple, wine-specific model |
Drawing the Markov Owl 🦉 / More complete model, stratify by Wine Origin, O_{X[i]} / Fit the wine origin model |
BONUS: Survival Analysis / Statistical Model / Finding reasonable hyperparameter for \alpha |
3. Statistical Models / Fitting Direct Causal Effect Model |
3. Statistical Models / Statistical models for admissions / Fitting Total Causal Effect Model |
Counts and Poisson Regression / Scientific model that includes innovation and technology loss / Determine good prior hyperparams |
Confounded Admissions / Direct Effect Estimator (now confounded due to common ability cause) / Fit the (confounded) Direct Effect Model |
Confounded Admissions / Sensitivity Analysis: Modeling latent ability confound variable / Fit the latent ability model |
Confounded Admissions / Total Effect Estimator / Fit the Total Effect Model |
Counts and Poisson Regression / Comparing Models / Model A - Global Intercept model |
Counts and Poisson Regression / Comparing Models / Model B - Interaction model |
BONUS: Simpons’s Pandora’s Box / Nonlinear Haunting / Partially Stratified Model – \text{logit}(p) = \alpha + \beta_{Z[i]} X_i |
BONUS: Simpons’s Pandora’s Box / Nonlinear Haunting / Try a fully-stratified model – \text{logit}(p_i) = \alpha_{Z[i]} + \beta_{Z[i]}X_i |
BONUS: Simpons’s Pandora’s Box / Nonlinear Haunting / Unstratified Model – \text{logit}(p_i) = \alpha + \beta X_i |
Ethics & Trolley Problem Studies / Ordered Monotonic Predictors / Assessing the Direct Effect of Education: Stratifying by Gender & Age |
Ethics & Trolley Problem Studies / What about competing causes? / Fit the gender-stratified model |
Ethics & Trolley Problem Studies / Statistical Model / Starting off easy |
Case Study: Reed Frog Survival / Building a Multilievel (Hierarchical) Model / Fit the multi-level model |
BONUS: Fixed Effects, Multilevel Models, & Mundlak Machines / Random Confounds / Fixed effect Model |
Case Study: Reed Frog Survival / Building a Multilievel (Hierarchical) Model / Comparing multi-level and fixed-sigma model / Fixed sigma model |
BONUS: Fixed Effects, Multilevel Models, & Mundlak Machines / Random Confounds / Multilevel Model |
Case Study: Reed Frog Survival / Including the presence of predators / Multilevel model with predator effects |
BONUS: Fixed Effects, Multilevel Models, & Mundlak Machines / Random Confounds / Naive Model |
BONUS: Fixed Effects, Multilevel Models, & Mundlak Machines / Random Confounds / Mundlak Machines / Statistical Model |
Case Study: Reed Frog Survival / Let’s build a (multi-level) model / What about the prior variance \sigma? |
BONUS: Fixed Effects, Multilevel Models, & Mundlak Machines / Latent Mundlak Machine (aka “Full Luxury Bayes”) / 2. Y sub-model |
Fertility & Behavior in Bangladesh / Varying districs + urban / Fit the district-urban model |
Fertility & Behavior in Bangladesh / Start simple: varying districts |
Adding Correlated Features / Model with correlated features / A couple of notes ⚠️ |
BONUS: Non-centered (aka Transformed) Priors / Example: Devil’s Funnel prior / Centered-prior models |
BONUS: Non-centered (aka Transformed) Priors / Non-centered prior |
Adding Correlated Features / Previous model – Using uncorrelated urban features |
What Motivates Sharing? / Including Predictive Household features / Add observed confounds variables to simulated dataset |
What Motivates Sharing? / 3) Statistical Model / Fit Wealth Gifting model on simulated data (validation) |
What Motivates Sharing? / 3) Statistical Model / Fitting the social ties model / Notes |
Phylogenetic Regression / Two equivalent formulations of Linear Regression / Classic Linear Regression |
Gaussian Processes (in the abstract) / Stratify by population size / Prior Predictive / Fit population-only model for comparison |
Phylogenetic Regression / From Model to Kernel / Fit the full Phylogentic model. |
Phylogenetic Regression / From Model to Kernel / Influence of Group Size on Brain Size |
Gaussian Processes (in the abstract) / Distance-based model / Model the data |
Gaussian Processes (in the abstract) / Stratify by population size / Prior Predictive |
Phylogenetic Regression / From Model to Kernel / Distance-only model / PyMC Implementation |
Phylogenetic Regression / Two equivalent formulations of Linear Regression / \text{MVNormal} Linear Regression |
Misclassification / Paternity in Himba pastoralist culture / Fit analogous model without accouting for misclassification |
Modeling Measurment / Compare causal effects for models that do and do not model measurement error / Fit model that considers no measurement error |
Modeling Measurment / Let’s start simpler / Fit the Divorce Measurement Error Model |
Modeling Measurment / Let’s start simpler / 4 Submodels / Marriage Rate Measurement Error Model |
Measurement Error / Myth: Measurement error can only decrease an effect, not increase |
Misclassification / Paternity in Himba pastoralist culture / Fit the model with misclassification error / Notes |
BONUS: Floating Point Monsters / Previous Paternity measurement model with log-scaling tricks |
Revisiting Phylogenetic Regression / Drawing the missing owl 🦉 / Complete case model for comparison |
Revisiting Phylogenetic Regression / Drawing the missing owl 🦉 / 3. Impute G using a G-specific submodels / 3. Fit model that combines phylogeny and M \rightarrow G |
Revisiting Phylogenetic Regression / Drawing the missing owl 🦉 / 2. Impute G,M naively, ignoring models for each / Fit the naive imputation model |
Revisiting Phylogenetic Regression / Drawing the missing owl 🦉 / 4. Impute B,G,M using a submodel for each |
Revisiting Phylogenetic Regression / Drawing the missing owl 🦉 / 3. Impute G using a G-specific submodels / 2. Model that only includes Social group phylogentic interactions |
Revisiting Phylogenetic Regression / Drawing the missing owl 🦉 / 3. Impute G using a G-specific submodels / 1. Model that only models effect of body mass on group size M \rightarrow G |
GLMs & GLMM & Generalized Linear Habits / Revisiting Modeling Human Height ⚖️ / Fit the dimensionless cylinder model |
GLMs & GLMM & Generalized Linear Habits / Revisiting Modeling Human Height ⚖️ / Statistical Model / Fit the statistical model |
Population Dynamics / Implement the statistical model / PyMC implementation details |
Choices, observation, and learning strategies 🟦🟥 / State-based Model / State-based statistical Model |
Bayesian Parametric Survival Analysis / Accelerated failure time models / Log-logistic survival regression |
Bayesian Parametric Survival Analysis / Accelerated failure time models / Weibull survival regression |
Censored Data Models / Censored data models / Model 1 - Imputed Censored Model of Censored Data |
Censored Data Models / Censored data models / Model 2 - Unimputed Censored Model of Censored Data |
Censored Data Models / Uncensored Model |
Frailty and Survival Regression Models / Accelerated Failure Time Models |
Frailty and Survival Regression Models / Fit Basic Cox Model with Fixed Effects |
Frailty and Survival Regression Models / Fit Model with Shared Frailty terms by Individual |
Bayesian Survival Analysis / Bayesian proportional hazards model |
Bayesian Survival Analysis / Bayesian proportional hazards model / Time varying effects |
Reparameterizing the Weibull Accelerated Failure Time Model / Parameterization 1 |
Reparameterizing the Weibull Accelerated Failure Time Model / Parameterization 2 |
Reparameterizing the Weibull Accelerated Failure Time Model / Parameterization 3 |
Analysis of An AR(1) Model in PyMC |
Analysis of An AR(1) Model in PyMC / Extension to AR(p) |
Air passengers - Prophet-like model / Part 1: linear trend |
Air passengers - Prophet-like model / Part 2: enter seasonality |
Inferring parameters of SDEs using a Euler-Maruyama scheme / Example Model |
Forecasting with Structural AR Timeseries / Complicating the Picture / Specifying a Trend Model |
Forecasting with Structural AR Timeseries / Specifying the Model |
Forecasting with Structural AR Timeseries / Complicating the picture further / Specifying the Trend + Seasonal Model |
Forecasting with Structural AR Timeseries / Complicating the Picture / Wrapping our model into a function |
Multivariate Gaussian Random Walk / Model |
Time Series Models Derived From a Generative Graph / Motivation / Define AR(2) Process / Posterior |
Bayesian Vector Autoregressive Models / Handling Multiple Lags and Different Dimensions |
Non-Linear Change Trajectories / A Minimal Model |
Non-Linear Change Trajectories / Adding in Polynomial Time |
Non-Linear Change Trajectories / Behaviour over time |
Non-Linear Change Trajectories / Comparing Trajectories across Gender |
Modelling Change over Time. / Model controlling for Peer Effects |
Modelling Change over Time. / The Unconditional Mean Model |
Modelling Change over Time. / The Uncontrolled Effects of Parental Alcoholism |
Modelling Change over Time. / Unconditional Growth Model |
Stochastic Volatility model / Fit Model |
GLM: Mini-batch ADVI on hierarchical regression model |
Empirical Approximation overview / 2d density |
Empirical Approximation overview / Multimodal density |
Pathfinder Variational Inference |
Introduction to Variational Inference with PyMC / Basic setup |
Introduction to Variational Inference with PyMC / Distributional Approximations |