Welcome to the documentation of “Debias-Infer”!

Debias-Infer is a Python library for conducting valid and efficient inference on high-dimensional linear models with missing outcomes.

A Preview into the Proposed Debiasing Inference Method

The proposed debiasing method introduces a novel debiased estimator for inferring the linear regression function with “missing at random (MAR)” outcomes. The key idea is to correct the bias of the Lasso solution [2] with complete-case data through a quadratic debiasing program with box constraints and construct the confidence interval based on the asymptotic normality of the debiased estimator.

More details can be found in Methodology and the reference paper [1].

Note

This project is under active development.

Contents:

• Installation Guide
  • Dependencies
  • Quick Start
• Methodology
  • Proposed Debiasing Inference Procedure
  • References
• Proposed Debiasing Method for Inferring High-Dimensional Linear Models (Example Code via Debias-Infer)
  • Simulate Random Sample
  • Detailed Procedures of Applying Our Proposed Debiasing Method
• API References
  • Function Gallery of the Debiasing Inference Method

References

[1]

Yikun Zhang, Alexander Giessing, Yen-Chi Chen (2023+) Efficient Inference on High-Dimensional Linear Models with Missing Outcomes.

[2]

Robert Tibshirani (1996). Regression Shrinkage and Selection via the Lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology 58, no.1: 267-288.