Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

Université de Montréal

$23,000.00

Dependence Modelling for Arbitrary Distributions

FY 2023-24

Summary
Dependence models are very important in several fields such as actuarial science, biostatistics, finance, and hydro-climatology to name a few. For example, in order to predict a variable of interest like the amount of precipitations during a given period, one has to take into account other variables (covariates) such as temperature, humidity, as well as previous values of precipitations. In order to predict accurately the precipitations in a near future, one should take into account the statistical dependence between the precipitations and the covariates. This can be done by using a dependence function called a copula. So far, most of the work done on copula models focused on the assumption that the variables have continuous distribution functions. However, for data in several fields of applications, this assumption is not met. Again, taking the example of precipitations, it is likely that there is no precipitation at all during several periods of time, so the assumption of a continuous distribution is not correct. In fact, in this case, the distribution is called zero-inflated, since it might be characterized by a continuous distribution with a point mass at zero. There are also many other examples of variables with non-continuous distributions, such as rounded observations, which usually have ties, and the obvious case when variables are discrete or even binary. All these cases are covered by what is called hereafter arbitrary distributions. In addition to the arbitrary distribution problem, there is also the problem of serial dependence to take into account, i.e., the current values of the variable of interest and covariates may depend on previous values. This problem of serial dependence within the framework of copulas has been considered again for continuous distributions. As I became more and more interested in the impact of climate change to public health, I realized that most data I would need to analyze have arbitrary distributions, display serial dependence and/or non-stationarity, and there are often a large number of variables to consider. Motivated by the data features mentioned above, my research program is composed of the following three themes: In the first theme, I will study inference for copulas for high-dimensional data with arbitrary distributions; in the second theme, I will work on copula-based dependence models and inference for serially dependent data with arbitrary distributions; in the last theme, I will study diagnostic tests for time series with arbitrary distributions. This research program will help researchers in several fields of applications to use statistical methods adapted to their data. For this purpose, I will develop open-access statistical computing programs (R packages) in order to make accessible the proposed statistical methodologies.
Program
Discovery Grants Program - Individual
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
RGPIN-2021-03013
Name nom
Nasri, Bouchra
Department département
Médecine sociale et préventive
Organization id
63
Institution établissement
Université de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2023
Competition year année de concours
2021
Award amount
23000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1100
Area of application group en
Health, education and social services
Area of application group fr
Santé publique, éducation et services sociaux
Area of application en
Health, education and social services
Area of application fr
Santé publique, éducation et services sociaux
Research subject code
9999
Research subject group en
Not available
Research subject group fr
Not available
Research subject en
Non disponible
Research subject fr
Non disponible
Application title
Dependence Modelling for Arbitrary Distributions
Keyword
Arbitrary distributions; Copulas; Goodness-of-fit; Semiparametric estimation; Time series
Application summary
Dependence models are very important in several fields such as actuarial science, biostatistics, finance, and hydro-climatology to name a few. For example, in order to predict a variable of interest like the amount of precipitations during a given period, one has to take into account other variables (covariates) such as temperature, humidity, as well as previous values of precipitations. In order to predict accurately the precipitations in a near future, one should take into account the statistical dependence between the precipitations and the covariates. This can be done by using a dependence function called a copula. So far, most of the work done on copula models focused on the assumption that the variables have continuous distribution functions. However, for data in several fields of applications, this assumption is not met. Again, taking the example of precipitations, it is likely that there is no precipitation at all during several periods of time, so the assumption of a continuous distribution is not correct. In fact, in this case, the distribution is called zero-inflated, since it might be characterized by a continuous distribution with a point mass at zero. There are also many other examples of variables with non-continuous distributions, such as rounded observations, which usually have ties, and the obvious case when variables are discrete or even binary. All these cases are covered by what is called hereafter arbitrary distributions. In addition to the arbitrary distribution problem, there is also the problem of serial dependence to take into account, i.e., the current values of the variable of interest and covariates may depend on previous values. This problem of serial dependence within the framework of copulas has been considered again for continuous distributions. As I became more and more interested in the impact of climate change to public health, I realized that most data I would need to analyze have arbitrary distributions, display serial dependence and/or non-stationarity, and there are often a large number of variables to consider. Motivated by the data features mentioned above, my research program is composed of the following three themes: In the first theme, I will study inference for copulas for high-dimensional data with arbitrary distributions; in the second theme, I will work on copula-based dependence models and inference for serially dependent data with arbitrary distributions; in the last theme, I will study diagnostic tests for time series with arbitrary distributions. This research program will help researchers in several fields of applications to use statistical methods adapted to their data. For this purpose, I will develop open-access statistical computing programs (R packages) in order to make accessible the proposed statistical methodologies.