StatsPAI
StatsPAI is an Agent-native Python library for causal inference and applied econometrics — unified API, broad cross-method coverage, structured result objects,
What is StatsPAI?
StatsPAI is an Agent-native Python library for causal inference and applied econometrics — unified API, broad cross-method coverage, structured result objects, machine-readable schemas, Skills, an MCP server, and R/Stata parity validation.
Overview of StatsPAI
What it is
StatsPAI is an Agent-native Python library for causal inference and applied econometrics. It provides a unified API across a broad range of estimators, structured result objects, machine-readable schemas, an MCP server, and R/Stata parity validation. It is designed to let empirical researchers use Stata and R workflows natively in Python.
Key features
- Unified API with methods mirroring Stata (e.g.,
regress,ivregress,reghdfe,csdid,rdrobust,synth,psmatch2,esttab) and R (e.g.,lm,fixest,did,rdrobust,Synth,DoubleML,MatchIt,modelsummary). - Stata-style conventions for standard errors (
vce=,robust=, cluster-variable inline) and small-sample factors. - Post-estimation methods:
.summary(),.tidy(),.plot(),.to_latex(),.to_docx(),.to_agent_summary(). - Agent-native access with machine-readable schemas (
sp.list_functions(),sp.describe_function(),sp.function_schema()) and a bundledstatspai-mcpserver. - Companion tooling for Stata translation and cross-checking, plus skill repos for empirical research workflows.
- Every public function carries a
validation_statusindicating certified/validated evidence versus API-stable breadth.
Use cases
Applied causal inference and econometric modeling in Python, including ordinary least squares (OLS), instrumental variables, high-dimensional fixed effects, staggered difference-in-differences, regression discontinuity, synthetic control, propensity score matching, double machine learning, and publication-ready table export.
Getting started
bash pip install statspai Python 3.10 – 3.13 on Linux and macOS. Optional extras include plotting (statspai[plotting]), fixest wrappers (statspai[fixest]), Bayesian estimators (statspai[bayes]), neural causal models (statspai[neural]), accelerated backends (statspai[performance]), Rust kernels (statspai[rust]), and spatial weights (statspai[spatial]).
License and pricing
License: MIT. Pricing: open source.
Based on the project's public README and repository metadata.