cbps 0.2.0__py3-none-any.whl

cbps/logging_config.py

@@ -0,0 +1,45 @@
+"""CBPS package logging configuration.
+
+Provides a package-level logger that users can control via standard
+Python logging mechanisms.
+
+Examples
+--------
+>>> from cbps import set_verbosity
+>>> set_verbosity(1)  # Enable INFO-level progress messages
+>>> set_verbosity(2)  # Enable DEBUG-level diagnostics
+>>> set_verbosity(0)  # Restore default (WARNING only)
+"""
+import logging
+
+# Package-level logger
+logger = logging.getLogger('cbps')
+logger.addHandler(logging.NullHandler())  # No output by default
+
+
+def set_verbosity(level: int = 0):
+    """Set CBPS package verbosity level.
+
+    Parameters
+    ----------
+    level : int
+        0 = WARNING only (default, production)
+        1 = INFO (progress messages)
+        2 = DEBUG (detailed diagnostics)
+    """
+    if level == 0:
+        logger.setLevel(logging.WARNING)
+    elif level == 1:
+        logger.setLevel(logging.INFO)
+    else:
+        logger.setLevel(logging.DEBUG)
+
+    # Add console handler if not already present
+    has_console = any(
+        isinstance(h, logging.StreamHandler) and not isinstance(h, logging.NullHandler)
+        for h in logger.handlers
+    )
+    if not has_console:
+        handler = logging.StreamHandler()
+        handler.setFormatter(logging.Formatter('[CBPS] %(levelname)s: %(message)s'))
+        logger.addHandler(handler)

cbps/msm/__init__.py

@@ -0,0 +1,45 @@
+"""
+Marginal Structural Models (MSM) Module.
+
+This module implements the Covariate Balancing Propensity Score (CBPS)
+methodology for marginal structural models, as developed by Imai and
+Ratkovic (2015). MSMs enable robust causal inference with time-varying
+treatments and confounders in longitudinal settings.
+
+Functions
+---------
+CBMSM
+    Formula-based interface for MSM weight estimation.
+cbmsm_fit
+    Matrix interface for advanced users.
+
+Classes
+-------
+CBMSMResults
+    Container for fitted weights, coefficients, and diagnostics.
+
+Notes
+-----
+The module estimates inverse probability weights by solving a GMM problem
+where moment conditions are derived from the covariate balancing property
+of MSM weights. Key features include:
+
+- Time-invariant or time-varying propensity score coefficients
+- Stabilized weights P(T)/P(T|X) for variance reduction
+- Low-rank covariance approximation for computational efficiency
+- Orthogonal moment conditions based on 2^J factorial design framework
+
+References
+----------
+Imai, K. and Ratkovic, M. (2015). Robust estimation of inverse probability
+weights for marginal structural models. Journal of the American Statistical
+Association, 110(511), 1013-1023. https://doi.org/10.1080/01621459.2014.956872
+"""
+
+from cbps.msm.cbmsm import CBMSM, cbmsm_fit, CBMSMResults
+
+__all__ = [
+    "CBMSM",
+    "cbmsm_fit",
+    "CBMSMResults",
+]