<?xml version="1.0" encoding="UTF-8"?><?xml-stylesheet type="text/xsl" href="static/style.xsl"?><OAI-PMH xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/ http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd"><responseDate>2026-11-22T11:38:19Z</responseDate><request verb="GetRecord" identifier="oai:uwspace.uwaterloo.ca:10012/3442" metadataPrefix="dim">https://dspacemainprd01.lib.uwaterloo.ca/server/oai/request</request><GetRecord><record><header><identifier>oai:uwspace.uwaterloo.ca:10012/3442</identifier><datestamp>2017-03-21T16:12:32Z</datestamp><setSpec>com_10012_9924</setSpec><setSpec>com_10012_1</setSpec><setSpec>col_10012_9934</setSpec><setSpec>col_10012_6</setSpec></header><metadata><dim:dim xmlns:dim="http://www.dspace.org/xmlns/dspace/dim" xmlns:doc="http://www.lyncode.com/xoai" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.dspace.org/xmlns/dspace/dim http://www.dspace.org/schema/dim.xsd">
   <dim:field mdschema="dc" element="contributor" qualifier="author" authority="69fb59f8-e841-4c95-9847-6026e04ea616" confidence="600">Li, Pengfei</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="accessioned">2007-12-14T19:12:26Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="available">2007-12-14T19:12:26Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="issued">2007-12-14T19:12:26Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="submitted">2007-12-11</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="uri">http://hdl.handle.net/10012/3442</dim:field>
   <dim:field mdschema="dc" element="description" qualifier="abstract" lang="en">Mixture models provide a natural framework for&#xd;
unobserved heterogeneity in a population.&#xd;
They are widely applied in astronomy, biology,&#xd;
engineering,  finance, genetics, medicine, social sciences,&#xd;
and other areas.&#xd;
&#xd;
An important first step for using mixture models is the test&#xd;
of homogeneity. Before one tries to fit a mixture model,&#xd;
it might be of value to know whether the data arise from a&#xd;
homogeneous or heterogeneous population. If the data are&#xd;
homogeneous, it is not even necessary to go into mixture modeling.&#xd;
The rejection of the homogeneous model may also have scientific implications.&#xd;
For example, in classical statistical genetics,&#xd;
it is often suspected that only a subgroup of patients have a&#xd;
disease gene which is linked to the marker. Detecting&#xd;
the existence of this subgroup amounts to the rejection of&#xd;
a homogeneous null model in favour of a two-component&#xd;
mixture model. This problem has attracted intensive&#xd;
research recently. This thesis makes substantial contributions&#xd;
in this area of research.&#xd;
&#xd;
&#xd;
Due to partial loss of identifiability, classic inference methods&#xd;
such as the likelihood ratio test (LRT) lose their usual elegant&#xd;
statistical properties. The limiting distribution of the LRT&#xd;
often involves complex Gaussian processes,&#xd;
which can be  hard to implement in data analysis.&#xd;
The modified likelihood ratio test (MLRT) is found to be a nice&#xd;
alternative of the LRT.  It restores the identifiability by introducing&#xd;
a penalty to the log-likelihood function.&#xd;
Under some mild conditions,&#xd;
the limiting distribution of the MLRT is&#xd;
1/2\chi^2_0+1/2\chi^2_1,&#xd;
where \chi^2_{0} is a point mass at 0.&#xd;
This limiting distribution is convenient to use in real data analysis.&#xd;
The choice of the penalty functions in the MLRT is very flexible.&#xd;
A good choice of the penalty enhances the power of the MLRT.&#xd;
In this thesis, we first introduce a new class of penalty functions,&#xd;
with which the MLRT enjoys a significantly improved power for testing&#xd;
homogeneity.&#xd;
&#xd;
&#xd;
The main contribution of this thesis is to propose a new class of&#xd;
methods for testing homogeneity. Most existing methods in the&#xd;
literature for testing of homogeneity, explicitly or implicitly, are&#xd;
derived under the condition of  finite Fisher information and a&#xd;
compactness assumption on the space of the mixing parameters. The&#xd;
finite Fisher information condition can prevent their usage to many&#xd;
important mixture models, such as the mixture of geometric&#xd;
distributions, the mixture of exponential distributions and more&#xd;
generally mixture models in scale distribution families. The&#xd;
compactness assumption often forces applicants to set artificial&#xd;
bounds for the parameters of interest and makes the resulting&#xd;
limiting distribution dependent on these bounds. Consequently,&#xd;
developing a method without such restrictions is a dream of many&#xd;
researchers. As it will be seen, the proposed EM-test in this thesis&#xd;
is free of these shortcomings.&#xd;
&#xd;
The EM-test combines the merits of the classic LRT and score test.&#xd;
The properties of the EM-test are particularly easy to investigate&#xd;
under single parameter mixture models.&#xd;
It has a simple limiting distribution&#xd;
0.5\chi^2_0+0.5\chi^2_1, the same as the MLRT.&#xd;
This result is applicable to mixture models without requiring&#xd;
the restrictive regularity conditions described earlier.&#xd;
&#xd;
The normal mixture model is a very popular model in applications.&#xd;
However it does not satisfy the strong identifiability condition,&#xd;
which imposes substantial technical difficulties in the study of the&#xd;
asymptotic properties. Most existing methods do not directly apply&#xd;
to the normal mixture models, so the asymptotic properties have to&#xd;
be developed separately. We investigate the use of the EM-test to&#xd;
normal mixture models and its limiting distributions are derived.&#xd;
For the homogeneity test in the presence of the structural&#xd;
parameter, the limiting distribution is a simple function of the&#xd;
0.5\chi^2_0+0.5\chi^2_1 and \chi^2_1 distributions. The test&#xd;
with this limiting distribution is still very convenient to&#xd;
implement. For normal mixtures in both mean and variance parameters,&#xd;
the limiting distribution of the EM-test is found be to \chi^2_2.&#xd;
&#xd;
Mixture models are also widely used in the analysis of the&#xd;
directional data. The von Mises distribution is often regarded as&#xd;
the circular normal model. Interestingly, it satisfies the strong&#xd;
identifiability condition and the parameter space of the mean&#xd;
direction is compact. However the theoretical results in the single&#xd;
parameter mixture models can not directly apply to the von Mises&#xd;
mixture models. Because of this, we also study the application of&#xd;
the EM-test to von Mises mixture models in the presence of the&#xd;
structural parameter. The limiting distribution of the EM-test is&#xd;
also found to be 0.5\chi^2_0+0.5\chi^2_1.&#xd;
&#xd;
Extensive simulation results are obtained to examine the precision&#xd;
of the approximation of the limiting distributions to the finite&#xd;
sample distributions of the EM-test. The type I errors with the&#xd;
critical values determined by the limiting distributions are found&#xd;
to be close to nominal values. In particular, we also propose&#xd;
several precision enhancing methods, which are found to work well.&#xd;
Real data examples are used to illustrate the use of the EM-test.</dim:field>
   <dim:field mdschema="dc" element="language" qualifier="iso" lang="en">en</dim:field>
   <dim:field mdschema="dc" element="publisher" lang="en">University of Waterloo</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Finite mixture model</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Homogeneity test</dim:field>
   <dim:field mdschema="dc" element="subject" qualifier="program" lang="en">Statistics</dim:field>
   <dim:field mdschema="dc" element="title" lang="en">Hypothesis Testing in Finite Mixture Models</dim:field>
   <dim:field mdschema="dc" element="type" lang="en">Doctoral Thesis</dim:field>
   <dim:field mdschema="dc" element="pending" lang="en">false</dim:field>
   <dim:field mdschema="uws-etd" element="degree" qualifier="department" lang="en">Statistics and Actuarial Science</dim:field>
   <dim:field mdschema="uws-etd" element="degree" lang="en">Doctor of Philosophy</dim:field>
   <dim:field mdschema="uws" element="typeOfResource" lang="en">Text</dim:field>
   <dim:field mdschema="uws" element="peerReviewStatus" lang="en">Unreviewed</dim:field>
   <dim:field mdschema="uws" element="scholarLevel" lang="en">Graduate</dim:field>
   <dim:field mdschema="others" element="access-status">open.access</dim:field>
</dim:dim>
</metadata></record></GetRecord></OAI-PMH>