Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

University of New Brunswick

$12,000.00

Methods for cluster analysis and inference for clustered count data

FY 2010-11

Summary
The proposed research program is aiming at two related directions: one is to develop methodology and algorithms for cluster analysis and the other is to conduct statistical inference for clustered count data in the presence of multiple sources of uncertainty. Cluster analysis seeks to classify objects into meaningful subgroups; it is widely used in many disciplines. My research in this area concerns primarily detecting linearly shaped clusters in a data set, which we shall refer to as ""linear clustering"". In a two-dimensional space, the task is to cluster data points around a few lines; in a higher-dimensional space, this involves clustering data points around some hyperplanes. In my earlier work, I developed a mixture model-based approach. This research will address issues of robustness, asymptotics and efficient computing. The objectives are a) to develop efficient and robust algorithms for cluster analysis and b) to understand the statistical properties of the algorithms in a theoretical context. The research will be disseminated by a statistical toolbox for cluster analysis and it has direct applications in genetic studies. Observations in count data take nonnegative integer values. In diverse disciplines, count data are clustered in a hierarchy. In addition, zero-inflation, overdispersion, measurement errors and missing data etc. pose further challenges to statistical inference. Hierarchical models are flexible and effective tools to tackle these challenges. Together with a colleague, I used compound Poisson distributions to account for zero-inflation in hierarchical models. The objectives of this research are a) to further study model building strategies for various clustered count data while incorporating multiple sources of uncertainty, b) to investigate model adequacy and computational efficiency and c) to make hierarchical models more accessible to the general practitioners of various disciplines.
Program
Discovery Grants Program - Individual
Location
New Brunswick, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
371505-2010
Name nom
Yan, Guohua
Department département
Mathematics and Statistics - Fredericton
Organization id
75
Institution établissement
University of New Brunswick
Province en
New Brunswick
Province fr
Nouveau-Brunswick
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2010
Competition year année de concours
2010
Award amount
12000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1211
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Medical and health sciences
Area of application fr
Sciences médicales et de la santé
Research subject code
3011
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Multivariate analysis
Research subject fr
Analyse multivariée
Partie
1
Nb partie
5
Application title
Methods for cluster analysis and inference for clustered count data
Keyword
NSERC
Application summary
The proposed research program is aiming at two related directions: one is to develop methodology and algorithms for cluster analysis and the other is to conduct statistical inference for clustered count data in the presence of multiple sources of uncertainty. Cluster analysis seeks to classify objects into meaningful subgroups; it is widely used in many disciplines. My research in this area concerns primarily detecting linearly shaped clusters in a data set, which we shall refer to as ""linear clustering"". In a two-dimensional space, the task is to cluster data points around a few lines; in a higher-dimensional space, this involves clustering data points around some hyperplanes. In my earlier work, I developed a mixture model-based approach. This research will address issues of robustness, asymptotics and efficient computing. The objectives are a) to develop efficient and robust algorithms for cluster analysis and b) to understand the statistical properties of the algorithms in a theoretical context. The research will be disseminated by a statistical toolbox for cluster analysis and it has direct applications in genetic studies. Observations in count data take nonnegative integer values. In diverse disciplines, count data are clustered in a hierarchy. In addition, zero-inflation, overdispersion, measurement errors and missing data etc. pose further challenges to statistical inference. Hierarchical models are flexible and effective tools to tackle these challenges. Together with a colleague, I used compound Poisson distributions to account for zero-inflation in hierarchical models. The objectives of this research are a) to further study model building strategies for various clustered count data while incorporating multiple sources of uncertainty, b) to investigate model adequacy and computational efficiency and c) to make hierarchical models more accessible to the general practitioners of various disciplines.
University of New Brunswick - $12,000 Grant (2010-11) | Canada Spends