Back to search

Cornell University

Functional Single Index Model and Jensen Effect

Abstract

dc:description.abstract

In natural ecosystems, environmental conditions are highly variable over time and space. There is much empirical evidence to show that environmental variability has significant effects on individuals and populations. The nonlinear response is modeled in this thesis in a functional single index model (FSIM), or its generalized version. In this thesis, we will investigate whether the impact is positive or negative. We will develop a nested nonlinear optimization algorithm with local quadratic approximation to estimate the second derivative of the curve in a FSIM. We will show convergence rates and consistency of the estimators. Even though our estimators perform very well theoretically, practical implementations require selections of initial value and bandwidth. Sometimes, our procedures are not able to select them properly, and will negatively impact estimation accuracy. Instead of estimating the curvature, we will estimate the Jensen Effect, or the sign of the Jensen's inequality, which only involves the link function, directly. Inspired by the SiZer method, we will skip the cross-validation step, but evaluate the Jensen Effect over a range of smoothing parameters. We will calculate a t-test statistics based on the Jensen Effect estimates, and perform a hypothesis test with critical values simulating from a Gaussian process. To analyze situations with variance heteroscedasticity in ecology data, we will consider logarithm of response variables, which lead to an exponential single index model. We will investigate the Jensen Effect, and propose a hypothesis test to see the impact of environmental factor. Then we will extend the methodology to generalized single index model for binary response, such as individual survival, or Poisson response, such as number of offspring.

Degree

thesis:*
Name thesis:degree_name
Ph.D., Statistics
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Statistics
Grantor
Cornell University
Year dc:date.issued
2019

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Ye, Zi
Chair dc:contributor.chair
  • Hooker, Giles J.
Committee members dc:contributor.committeemember
  • Wells, Martin Timothy
  • Ellner, Stephen Paul

Subjects

dc:subject × 3

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Dc Identifier Bibid
11050601
Dc Identifier Other
ProQuest Submission ID: 11500
ProQuest Publication ID: 13898294

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Ye, Zi. Functional Single Index Model and Jensen Effect. Doctor of Philosophy thesis, Cornell University, 2019. https://hdl.handle.net/1813/67618