PyMixEF 0.1 documentation
Mixed-effects modeling in Python
Specify, fit, diagnose, and preserve mixed-effects analyses with an inspectable model representation. Follow a statistical workflow, reproduce complete scientific examples, or look up the public Python API.
Start with your task¶
Choose an installation profile, fit a first model, and learn the declare → validate → compile → fit → diagnose workflow.
Map a scientific question to LMM, GLMM, MMRM, PK/ODE simulation, or a pharmacometric model declaration.
Ten pre-executed, assertion-backed case studies: three materials/catalysis analyses and seven biomedical/pharmaceutical workflows.
Task-oriented navigation plus generated, signature-level documentation for the complete public Python surface.
Analysis areas¶
Continuous outcomes
Gaussian ML/REML, multiple random blocks, conditional and population predictions, residual diagnostics, simulation, and reproducible results.
Discrete outcomes
Bernoulli, binomial, Poisson, and negative-binomial-2 reference fits using the explicitly labeled first-order Laplace path.
Repeated measures
Ordered residual covariance, missing-response auditing, longitudinal contrasts, and explicitly labeled degrees-of-freedom calculations.
Pharmacology
Canonical dosing events, closed-form one- and two-compartment PK, event-aware ODE integration, sensitivities, and residual-error models.
Inspectable models
Formula and typed DSL authoring compile to a backend-neutral, immutable, schema-versioned ModelIR with deterministic semantic hashes.
Reproducibility
Structured convergence, diagnostics, archives, manifests, validation bundles, comparison reports, and conservative external-format compatibility reports.
Search¶
Open search with Ctrl/⌘+K or /.
Search accepts concepts such as “conditional prediction,” exact symbols such as
Model.compile, warning codes, CLI commands, and tutorial terms. When you open
a result, every matching phrase is highlighted on the page; use the match bar
or Enter/Shift+Enter to move between hits.