Topics in statistical prediction and inference
Abstract/Contents
- Abstract
- In the era of big data with increasingly complex learning systems, reliable and valid statistical inference is crucial for extracting meaningful insights and ensuring the integrity of data analysis. For example, in error estimation and model evaluation, statistical inference provides principled methods to estimate and quantify uncertainty in prediction errors. In causal inference, randomized experiments and hypothesis testing are widely used to detect significant average treatment effect, etc. However, larger datasets and more complex settings can lead to misuse of existing inference tools in situations beyond their assumptions. This thesis explores the following challenges faced by modern statistical inference for the targets of prediction error and average treatment effect. For inference of prediction error, covariate shift may occur and time correlation could result in failure of CLT-based naive confidence intervals. For Inference of average treatment effect, "hunting" for significance can inflate type-I errors and false discovery rates in large-scale sequential experiments, and interference poses challenges to valid inference of exposure contrasts and optimal experimental design. In this thesis, we will examine particular settings with the above challenges and build upon existing literature to discuss principle-inspired methodologies to tackle them.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2023; ©2023 |
Publication date | 2023; 2023 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Xu, Hui |
---|---|
Degree supervisor | Tibshirani, Robert |
Thesis advisor | Tibshirani, Robert |
Thesis advisor | Hastie, Trevor |
Thesis advisor | Owen, Art |
Degree committee member | Hastie, Trevor |
Degree committee member | Owen, Art |
Associated with | Stanford University, School of Humanities and Sciences |
Associated with | Stanford University, Department of Statistics |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Hui Xu. |
---|---|
Note | Submitted to the Department of Statistics. |
Thesis | Thesis Ph.D. Stanford University 2023. |
Location | https://purl.stanford.edu/dx010jm9786 |
Access conditions
- Copyright
- © 2023 by Hui Xu
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...