國立虎尾科技大學 |

Causal Inference with Non-Standard Experimental Designs.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Causal Inference with Non-Standard Experimental Designs./
Author:	Wu, Han.
Description:	1 online resource (165 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-04, Section: A.
Contained By:	Dissertations Abstracts International85-04A.
Subject:	Design. -
Online resource:	click for full text (PQDT)
ISBN:	9798380481496

Causal Inference with Non-Standard Experimental Designs.
Wu, Han.

Causal Inference with Non-Standard Experimental Designs. - 1 online resource (165 pages)

Source: Dissertations Abstracts International, Volume: 85-04, Section: A.

Thesis (Ph.D.)--Stanford University, 2023.

Includes bibliographical references

The past decades have seen a comprehensive body of research dedicated to causal inference in conventional experimental designs. However, as technological innovations continue to foster a rapid influx of data across numerous fields, the datasets derived often exhibit new structures that stem from unconventional designs. The thesis at hand is centered around the development of methods for conducting causal inference, particularly when the design deviates from the standard, thereby making conventional methods inapplicable.Chapter 2 delves into the regression discontinuity design in cases where the running variable is a noisy measurement of a latent variable. We propose a novel design-based approach for estimation and inference. This approach proves effective when applied to a broad array of widely-used estimands.Chapter 3 explores adaptive experimentation in the context of delayed feedback. In subchapter 3.1, we extend Thompson sampling to the proportional hazard model and develop a method capable of overcoming challenges associated with vaccine trials. Subsequently, in subchapter 3.2, we study the behavior of Thompson sampling when delays are unrestricted, providing theoretical regret bounds and conducting extensive experiments.Chapter 4 investigates policy learning in scenarios involving multiple treatments or multiple outcomes. In subchapter 4.1, we propose methods for evaluating policies when cost constraints accompany multiple treatments. In subchapter 4.2, we introduce a personalized experimentation system that can learn interpretable policies from experimental data and is scalable for big datasets.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798380481496Subjects--Topical Terms:

595500
Design.
Index Terms--Genre/Form:

554714
Electronic books.

Causal Inference with Non-Standard Experimental Designs.
LDR:02816ntm a22003257 4500 001 1149252
005 20241007100331.5
006 m o d
007 cr bn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798380481496
035 $a (MiAaPQ)AAI30614627
035 $a (MiAaPQ)STANFORDzn121zh7709
035 $a AAI30614627
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Wu, Han. $3 1475417
245 1 0 $a Causal Inference with Non-Standard Experimental Designs.
264 0 $c 2023
300 $a 1 online resource (165 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-04, Section: A.
500 $a Advisor: Wager, Stefan.
502 $a Thesis (Ph.D.)--Stanford University, 2023.
504 $a Includes bibliographical references
520 $a The past decades have seen a comprehensive body of research dedicated to causal inference in conventional experimental designs. However, as technological innovations continue to foster a rapid influx of data across numerous fields, the datasets derived often exhibit new structures that stem from unconventional designs. The thesis at hand is centered around the development of methods for conducting causal inference, particularly when the design deviates from the standard, thereby making conventional methods inapplicable.Chapter 2 delves into the regression discontinuity design in cases where the running variable is a noisy measurement of a latent variable. We propose a novel design-based approach for estimation and inference. This approach proves effective when applied to a broad array of widely-used estimands.Chapter 3 explores adaptive experimentation in the context of delayed feedback. In subchapter 3.1, we extend Thompson sampling to the proportional hazard model and develop a method capable of overcoming challenges associated with vaccine trials. Subsequently, in subchapter 3.2, we study the behavior of Thompson sampling when delays are unrestricted, providing theoretical regret bounds and conducting extensive experiments.Chapter 4 investigates policy learning in scenarios involving multiple treatments or multiple outcomes. In subchapter 4.1, we propose methods for evaluating policies when cost constraints accompany multiple treatments. In subchapter 4.2, we introduce a personalized experimentation system that can learn interpretable policies from experimental data and is scalable for big datasets.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Design. $3 595500
650 4 $a Probability. $3 667790
650 4 $a Vaccines. $3 582184
650 4 $a Meals. $3 1467121
650 4 $a Feedback. $3 1047630
655 7 $a Electronic books. $2 local $3 554714
690 $a 0389
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a Stanford University. $3 1184533
773 0 $t Dissertations Abstracts International $g 85-04A.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30614627 $z click for full text (PQDT)