東華大學圖書館 |

Online Testing and Semiparametric Estimation of Complex Treatment Effects.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Online Testing and Semiparametric Estimation of Complex Treatment Effects./
作者:	Yu, Miao.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2021,
面頁冊數:	100 p.
附註:	Source: Dissertations Abstracts International, Volume: 83-05, Section: B.
Contained By:	Dissertations Abstracts International83-05B.
標題:	Sample size. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28688450
ISBN:	9798544205326

Online Testing and Semiparametric Estimation of Complex Treatment Effects.
Yu, Miao.

Online Testing and Semiparametric Estimation of Complex Treatment Effects. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 100 p.

Source: Dissertations Abstracts International, Volume: 83-05, Section: B.

Thesis (Ph.D.)--North Carolina State University, 2021.

This item must not be sold to any third party vendors.

A/B testing plays a critical role in the high-tech industry to guide product development and accelerate innovation. It performs formal null hypothesis statistical testing to determine if there is a significant difference between the two variants on the metric of interest. However, a typical A/B test presents two problems: (i) a fixed-horizon framework inflates the type I error under continuous monitoring; (ii) the homogeneous effects assumption fails to detect the heterogeneous treatment effects among the population. The first two parts of this thesis aim to develop an online test that can address these two problems simultaneously.In the first part, we propose an online test, named sequential score test (SST), to detect the multidimensional heterogeneous treatment effect under a generalized linear model, which is able to control the type I error under continuous monitoring. We provide an online p-value calculation for SST, making it convenient for continuous monitoring, and extend our tests to online multiple testing settings by controlling the false discovery rate. We examine the empirical performance of the proposed tests and compare them with a state-of-art online test, named mSPRT using simulations and a real dataset. The results show that our proposed test controls type I error at any time, has higher detection power, and allows quick inference on online A/B testing.In the second part, we propose an online test for subgroup treatment effects based on value difference, named SUBTLE. The proposed test allows the experimenters to "peek" the results during the experiment without harming the statistical guarantees. It considers a nonparametric model and aims to test if some subgroup of the population will benefit from the investigative treatment. If the testing result indicates the existence of such a subgroup, a subgroup will be identified using a readily available estimated optimal treatment rule. The empirical performance of our proposed test is examined on simulations and a real dataset. The results show that the SUBTLE has high detection power with controlled type I error at any time, is robust to noise covariates, and achieves early stopping compared with the corresponding fixed-horizon test.In the last part, we shift the attention to the estimation of time-dependent treatment effects with zero-inflated outcomes. Motivated by freemium mobile game data, we propose a class of multiplicative structural nested mean models for zero-inflated nonnegative outcomes, which flexibly describes the joint effect of a sequence of treatments in the presence of time-varying confounders. The proposed estimator solves a doubly robust estimating equation, where the nuisance functions, propensity score and conditional outcome means given confounders, are estimated parametrically or nonparametrically. To improve the accuracy, we leverage the characteristic of zero-inflated outcomes by estimating the conditional means in two parts, that is, separately modeling the probability of having positive outcomes given confounders and the mean outcome conditional on its being positive and confounders. We show that the proposed estimator is consistent and asymptotically normal as either the sample size or the follow-up time goes to infinity. Moreover, the typical sandwich formula can be used to estimate the variance of treatment effect estimators consistently, without accounting for the variation due to estimating nuisance functions. Simulation studies and an application to a freemium mobile game dataset are presented to demonstrate the empirical performance of the proposed method and support our theoretical findings.

ISBN: 9798544205326Subjects--Topical Terms:

3642155
Sample size.

Online Testing and Semiparametric Estimation of Complex Treatment Effects.
LDR:04738nmm a2200325 4500 001 2342400
005 20220318093134.5
008 241004s2021 ||||||||||||||||| ||eng d
020 $a 9798544205326
035 $a (MiAaPQ)AAI28688450
035 $a (MiAaPQ)NCState_Univ18402038861
035 $a AAI28688450
040 $a MiAaPQ $c MiAaPQ
100 1 $a Yu, Miao. $3 3181683
245 1 0 $a Online Testing and Semiparametric Estimation of Complex Treatment Effects.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 100 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-05, Section: B.
500 $a Advisor: Yang, Shu ;Song, Rui;Tang, Minh ;Lu, Wenbin.
502 $a Thesis (Ph.D.)--North Carolina State University, 2021.
506 $a This item must not be sold to any third party vendors.
520 $a A/B testing plays a critical role in the high-tech industry to guide product development and accelerate innovation. It performs formal null hypothesis statistical testing to determine if there is a significant difference between the two variants on the metric of interest. However, a typical A/B test presents two problems: (i) a fixed-horizon framework inflates the type I error under continuous monitoring; (ii) the homogeneous effects assumption fails to detect the heterogeneous treatment effects among the population. The first two parts of this thesis aim to develop an online test that can address these two problems simultaneously.In the first part, we propose an online test, named sequential score test (SST), to detect the multidimensional heterogeneous treatment effect under a generalized linear model, which is able to control the type I error under continuous monitoring. We provide an online p-value calculation for SST, making it convenient for continuous monitoring, and extend our tests to online multiple testing settings by controlling the false discovery rate. We examine the empirical performance of the proposed tests and compare them with a state-of-art online test, named mSPRT using simulations and a real dataset. The results show that our proposed test controls type I error at any time, has higher detection power, and allows quick inference on online A/B testing.In the second part, we propose an online test for subgroup treatment effects based on value difference, named SUBTLE. The proposed test allows the experimenters to "peek" the results during the experiment without harming the statistical guarantees. It considers a nonparametric model and aims to test if some subgroup of the population will benefit from the investigative treatment. If the testing result indicates the existence of such a subgroup, a subgroup will be identified using a readily available estimated optimal treatment rule. The empirical performance of our proposed test is examined on simulations and a real dataset. The results show that the SUBTLE has high detection power with controlled type I error at any time, is robust to noise covariates, and achieves early stopping compared with the corresponding fixed-horizon test.In the last part, we shift the attention to the estimation of time-dependent treatment effects with zero-inflated outcomes. Motivated by freemium mobile game data, we propose a class of multiplicative structural nested mean models for zero-inflated nonnegative outcomes, which flexibly describes the joint effect of a sequence of treatments in the presence of time-varying confounders. The proposed estimator solves a doubly robust estimating equation, where the nuisance functions, propensity score and conditional outcome means given confounders, are estimated parametrically or nonparametrically. To improve the accuracy, we leverage the characteristic of zero-inflated outcomes by estimating the conditional means in two parts, that is, separately modeling the probability of having positive outcomes given confounders and the mean outcome conditional on its being positive and confounders. We show that the proposed estimator is consistent and asymptotically normal as either the sample size or the follow-up time goes to infinity. Moreover, the typical sandwich formula can be used to estimate the variance of treatment effect estimators consistently, without accounting for the variation due to estimating nuisance functions. Simulation studies and an application to a freemium mobile game dataset are presented to demonstrate the empirical performance of the proposed method and support our theoretical findings.
590 $a School code: 0155.
650 4 $a Sample size. $3 3642155
650 4 $a Computer & video games. $3 3548317
650 4 $a Algorithms. $3 536374
650 4 $a Computer science. $3 523869
650 4 $a Recreation. $3 535376
690 $a 0338
690 $a 0984
690 $a 0814
710 2 $a North Carolina State University. $3 1018772
773 0 $t Dissertations Abstracts International $g 83-05B.
790 $a 0155
791 $a Ph.D.
792 $a 2021
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28688450