Valid Simultaneous Inference in High-Dimensional Settings (with the hdm\n package for R)

2018/09/13 by Philipp Bach, Victor Chernozhukov, Bach, Philipp +3
Mathematics · #Statistical Methods and Inference #Advanced Causal Inference Techniques

paper · pdf · doi:10.48550/arxiv.1809.04951

Abstract

Due to the increasing availability of high-dimensional empirical applications\nin many research disciplines, valid simultaneous inference becomes more and\nmore important. For instance, high-dimensional settings might arise in economic\nstudies due to very rich data sets with many potential covariates or in the\nanalysis of treatment heterogeneities. Also the evaluation of potentially more\ncomplicated (non-linear) functional forms of the regression relationship leads\nto many potential variables for which simultaneous inferential statements might\nbe of interest. Here we provide a review of classical and modern methods for\nsimultaneous inference in (high-dimensional) settings and illustrate their use\nby a case study using the R package hdm. The R package hdm implements valid\njoint powerful and efficient hypothesis tests for a potentially large number of\ncoeffcients as well as the construction of simultaneous confidence intervals\nand, therefore, provides useful methods to perform valid post-selection\ninference based on the LASSO.\n

Related

• Statistical Inference and Large-scale Multiple Testing for High-dimensional Regression Models
• High-Dimensional Metrics in R
• High-dimensional simultaneous inference with the bootstrap
• Inference in high-dimensional regression models without the exact or L^p sparsity
• Two-directional simultaneous inference for high-dimensional models