Targeted Learning Library
Python package for targeted inference.
targeted provides a number of methods for semi-parametric
estimation. The library also contains implementations of various
parametric models (including different discrete choice models) and
model diagnostics tools.
The implemention currently includes
- Risk regression models with binary exposure
(Richardson et al., 2017, doi:10.1080/01621459.2016.1192546)
- Augmented Inverse Probability Weighted estimators for missing
data and causal inference (Bang and Robins, 2005,
doi:10.1111/j.1541-0420.2005.00377.x)
- Model diagnostics based on cumulative residuals methods
- Efficient weighted Pooled Adjacent Violator Algorithms
- Nested multinomial logit models
Documentation and tutorials can be found at https://targetlib.org/python/.