Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

University of Calgary

$13,000.00

New Statistical Methodologies for the Analyses of Complex and Imperfect Functional Data

EF 2015-16

Résumé
This proposal represents my interest in developing new statistical methodologies to handle functional data analysis with complex data structure and imperfect data quality. The topics mentioned here arise from newly occurred data forms. For example, advanced monitor devices can simultaneously provide continuous, binomial and count observations at varying time point. Therefore, we obtain multivariate functional data mixed with different types of response variables. The functional data can also be collected by days within weeks and weeks within subjects, and thus they have multilevel property. Moreover, missing data may occur in the records due to varied reasons. The applicant has proposed an ambitious schedule to develop new statistical methodologies to handle these problems, which involves three short-term topics and two long-term topics.  <span style=""line-height: 1.5em;"">In short-term aim 1, the applicant will develop functional data approaches for multiple outcomes mixed with different types of response variables. One typical difficulty in this project is expensive computational cost. The applicant proposes to use a penalized quasilikelihood type approximation to unify different types of response variables into a ""pseudo"" normal distributed data and develops new algorithm to handle this problem. In short-term aim 2, the applicant will provide a flexible modeling strategy for three or higher level functional data. The applicant proposes to use dimension reduction approach for model simplicity and ease explanation. A new algorithm will be developed to incorporate both modeling and dimension reduction issues. In short-term aim 3, the applicant will develop new approaches for data incompleteness in functional data analysis. The applicant will solve the problem by using a particular bias-correct likelihood formulation for missingness-prone observations. The new likelihood has a conditional form, where improved computation algorithm will be developed to implement the inference.  The long-term research will study ""complex"" data structure with spatially correlation, and ""imperfect"" data quality with outcomes to have misclassification. The research will: (i) use a spatial correlation function in modeling to postulate the influence of spatial distance between subunits; (ii) consider binary outcomes with biased measurement, include the probabilities of correct classification into model estimation, and probably use the approximation approach studied in short-term aim 1 to take computation.  <span style=""line-height: 1.5em;"">The research proposal is expected to bring important changes in functional data analysis. Advanced methods will be developed to handle complex structures and imperfect data quality in various forms. Efficient algorithms will also be provided to handle issues in computations. Finally, undergraduate and graduate students will have opportunities to get statistical training in this project.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Alberta, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
RGPIN-2015-04409
Name nom
Li, Haocheng
Department département
Oncology - Oncology
Organization id
11
Institution établissement
University of Calgary
Province en
Alberta
Province fr
Alberta
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2015
Competition year année de concours
2015
Award amount
13000
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
1205
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Mathematical sciences
Area of application fr
Sciences mathématiques
Research subject code
3010
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Nonparametric inference
Research subject fr
Inférence non paramétrique
Partie
1
Nb partie
5
Application title
New Statistical Methodologies for the Analyses of Complex and Imperfect Functional Data
Keyword
NSERC
Application summary
<p>This proposal represents my interest in developing new statistical methodologies to handle functional data analysis with complex data structure and imperfect data quality. The topics mentioned here arise from newly occurred data forms. For example, advanced monitor devices can simultaneously provide continuous, binomial and count observations at varying time point. Therefore, we obtain multivariate functional data mixed with different types of response variables. The functional data can also be collected by days within weeks and weeks within subjects, and thus they have multilevel property. Moreover, missing data may occur in the records due to varied reasons. The applicant has proposed an ambitious schedule to develop new statistical methodologies to handle these problems, which involves three short-term topics and two long-term topics. </p> <p><span style=""line-height: 1.5em;"">In short-term aim 1, the applicant will develop functional data approaches for multiple outcomes mixed with different types of response variables. One typical difficulty in this project is expensive computational cost. The applicant proposes to use a penalized quasilikelihood type approximation to unify different types of response variables into a ""pseudo"" normal distributed data and develops new algorithm to handle this problem.</span></p> <p> In short-term aim 2, the applicant will provide a flexible modeling strategy for three or higher level functional data. The applicant proposes to use dimension reduction approach for model simplicity and ease explanation. A new algorithm will be developed to incorporate both modeling and dimension reduction issues.</p> <p> In short-term aim 3, the applicant will develop new approaches for data incompleteness in functional data analysis. The applicant will solve the problem by using a particular bias-correct likelihood formulation for missingness-prone observations. The new likelihood has a conditional form, where improved computation algorithm will be developed to implement the inference. </p> <p> The long-term research will study ""complex"" data structure with spatially correlation, and ""imperfect"" data quality with outcomes to have misclassification. The research will: (i) use a spatial correlation function in modeling to postulate the influence of spatial distance between subunits; (ii) consider binary outcomes with biased measurement, include the probabilities of correct classification into model estimation, and probably use the approximation approach studied in short-term aim 1 to take computation. </p> <p><span style=""line-height: 1.5em;"">The research proposal is expected to bring important changes in functional data analysis. Advanced methods will be developed to handle complex structures and imperfect data quality in various forms. Efficient algorithms will also be provided to handle issues in computations. Finally, undergraduate and graduate students will have opportunities to get statistical training in this project.</span></p>