RKHSMetaMod: An R Package to Estimate the Hoeffding Decomposition of a Complex Model by Solving RKHS Ridge Group Sparse Optimization Problem
Abstract:
In this paper, we propose an R package, called RKHSMetaMod, that implements a procedure for estimating a meta-model of a complex model. The meta-model approximates the Hoeffding decomposition of the complex model and allows us to perform sensitivity analysis on it. It belongs to a reproducing kernel Hilbert space that is constructed as a direct sum of Hilbert spaces. The estimator of the meta-model is the solution of a penalized empirical least-squares minimization with the sum of the Hilbert norm and the empirical -norm. This procedure, called RKHS ridge group sparse, allows both to select and estimate the terms in the Hoeffding decomposition, and therefore, to select and estimate the Sobol indices that are non-zero. The RKHSMetaMod package provides an interface from the R statistical computing environment to the C++ libraries Eigen and GSL. In order to speed up the execution time and optimize the storage memory, except for a function that is written in R, all of the functions of this package are written using the efficient C++ libraries through RcppEigen and RcppGSL packages. These functions are then interfaced in the R environment in order to propose a user-friendly package.
Consider a phenomenon described by a model depending on input
variables . This model from to
may be a known model that is calculable in all points of
, i.e. , or it may be an unknown regression model defined as
follows:
where the error is assumed to be centered with a finite
variance, i.e. and . The
components of are independent with a known law
on , a subset of .
The number of components of may be large. The model may
present high complexity as strong non-linearities and high order
interaction effects, and it is assumed to be square-integrable, i.e.
. Based on the data points
, we estimate a meta-model that approximates the
Hoeffding decomposition of . This meta-model belongs to a reproducing
kernel Hilbert space (RKHS), which is constructed as a direct sum of the
Hilbert spaces leading to an additive decomposition including variables
and interactions between them (Durrande et al. 2013). The estimator of the
meta-model is calculated by minimizing an empirical least-squares
criterion penalized by the sum of two penalty terms: the Hilbert norm
and the empirical norm (Huet and Taupin 2017). This procedure allows us to
select the subsets of variables that contribute to predict
. The estimated meta-model is used to perform sensitivity analysis,
and therefore, to determine the influence of each variable and groups of
them on the output variable .
In the classical framework of the sensitivity analysis, the function
is calculable in all points of , and one may use the method of
Sobol (1993) to perform the sensitivity analysis on . Let
us briefly introduce this method:
The independency between the components of leads to write the
function according to its Hoeffding decomposition
(Sobol 1993; Van der Vaart 1998):
The terms in this decomposition are defined using the conditional
expected values:
These terms are known as the constant term, main effects, interactions
of order two and so on. Let be the set of all subsets of
with dimension to . For all and
, let be the vector with components ,
. For a set let be its cardinality, and for
all , let
be the function
associated with in Equation ((2)). Then Equation
((2)) can be expressed as follows:
This decomposition is unique, all terms , are
centered, and they are orthogonal with respect to
. The functions and ,
in Equation ((3)) are square-integrable. As any two terms
of decomposition ((3)) are orthogonal, by squaring
((3)) and integrating it with respect to the distribution
of , a decomposition of the variance of is obtained as
follows:
The Sobol indices associated with the group of variables ,
are defined by:
For each , the expresses the fraction of the variance of
explained by . For all , when ,
the s are referred to as the first order indices; when
, i.e. and , they are referred
to as the second order indices or the interaction indices of order two
(between and ); and the same holds for .
The total number of the Sobol indices to be calculated is equal to
, which raises exponentially with the
number of the input variables . When is large, the evaluation of
all the indices can be computationally demanding and even not reachable.
In practice, only the indices of order not higher than two are
calculated. However, only the first and second order indices may not
provide a good information on the model sensitivities. In order to
provide better information on the model sensitivities, Homma and Saltelli (1996)
proposed to calculate the first order and the total indices defined as
follows:
Let be the set of all the subsets of
including , then .
For all , the denotes the total effect of
. It expresses the fraction of variance of explained by
alone and all the interactions of it with the other variables. The
total indices allow us to rank the input variables with respect to the
amount of their effect on the output variable. However, they do not
provide complete information on the model sensitivities as do all the
Sobol indices.
The classical computation of the Sobol indices is based on the Monte
Carlo methods (see for example: Sobol (1993) for the main
effect and interaction indices, and Saltelli (2002) for the main effect
and total indices). For models that are expensive to evaluate, the Monte
Carlo methods lead to a high computational burden. Moreover, in the case
where is large, is complex and the calculation of the variances
(see Equation ((4))) is numerically complicated or
not possible (as in the case where the model is unknown) the methods
described above are not applicable. Another approach consists in
approximating by a simplified model, called a meta-model, which is
much faster to evaluate and to perform sensitivity analysis on it.
Beside the approximations of the Sobol indices of at a lower
computational cost, a meta-model provides a deeper view of the input
variables effects on the model output. Among the meta-modelling methods
proposed in the literature, the expansion based on the polynomial Chaos
(Wiener 1938; Schoutens 2000) can be used to approximate
the Hoeffding decomposition of (Sudret 2008). The principle of
the polynomial Chaos is to project onto a basis of orthonormal
polynomials. The polynomial Chaos expansion of is written as
(Soize and Ghanem 2004):
where are the coefficients, and
are the multivariate orthonormal polynomials
associated with which are determined according to the distribution
of the components of . In practice, expansion ((6)) shall
be truncated for computational purposes, and the model may be
approximated by , where is
determined using a truncation scheme. The Sobol indices are obtained
then by summing up the squares of the suitable coefficients.
Blatman and Sudret (2011) proposed a method for truncating the polynomial Chaos
expansion and an algorithm based on the least angle regression for
selecting the terms in the expansion. In this approach, according to the
distribution of the components of , a unique family of orthonormal
polynomials is determined. However, this
family may not be necessarily the best functional basis to approximate
well.
Gaussian Process (GP) can also be used to construct meta-models as
highlighted in Welch et al. (1992), Oakley and O’Hagan (2004),
Kleijnen (2007, 2009), Marrel et al. (2009),
Durrande et al. (2012), and Le Gratiet et al. (2014). The principle is to consider
that the prior knowledge about the function , can be modelled by a
GP with a mean and a covariance
kernel . To perform sensitivity analysis from a
GP model, one may replace the model with the mean of the
conditional GP and deduce the Sobol indices from it. A review on the
meta-modelling based on the polynomial Chaos and the GP is presented in
Le Gratiet et al. (2017).
Durrande et al. (2013) considered a class of the functional approximation
methods similar to the GP and obtained a meta-model that satisfies the
properties of the Hoeffding decomposition. They proposed to approximate
by functions belonging to a RKHS which is a direct sum
of the Hilbert spaces. Their RKHS is constructed in a way
that the projection of onto , denoted , is an
approximation of the Hoeffding decomposition of . The function
is defined as the minimizer over the functions of the
criterion
Let be the scalar product in
, let also and be the reproducing kernels
associated with the RKHS and the RKHS
respectively. The properties of the RKHS insures that any
function ,
is written as
the following decomposition:
where is constant, and
is defined by
. For more details on
the RKHS construction and the definition of the Hilbert norm see Section
"RKHS construction" in the Appendix (supplementary materials).
For all , the functions are centered and for
all , the functions and are
orthogonal with respect to . Therefore, the
decomposition of the function presented in Equation
((7)) is its Hoeffding decomposition. As the function
belongs to the RKHS , it is decomposed as its
Hoeffding decomposition, , and
each function approximates the function in Equation
((3)). The number of the terms that should be
estimated in the Hoeffding decomposition of is equal to
, which may be huge since it rises very
quickly by increasing . In order to deal with this problem, in the
regression framework, one may estimate by a sparse meta-model
. To this end, the estimation of is
done on the basis of observations by minimizing a least-squares
criterion suitably penalized in order to deal with both the
non-parametric nature of the problem and the possibly large number of
functions that have to be estimated. In the classical framework of the
sensitivity analysis one may calculate a sparse approximation of
using least-squares penalized criterion as it is done in the
non-parametric regression framework. In order to obtain a sparse
solution of a minimization problem, the penalty function should enforce
the sparsity. There exists various ways of enforcing sparsity for a
minimization (maximization) problem, see for example
Hastie et al. (2015) for a review. Some methods, such as the Sparse
Additive Models (SpAM) procedure (Liu et al. 2009; Ravikumar et al. 2009) are based
on a combination of the -norm with the empirical -norm:
where
is the squared
empirical -norm of the univariate function . The Component
Selection and Smoothing Operator (COSSO) method developed by Lin and Zhang (2006)
enforces sparsity using a combination of the -norm with the Hilbert
norm:
.
Instead of focusing on only one penalty term, one may consider a more
general family of estimators, called the doubly penalized estimator,
which is obtained by minimizing a criterion penalized by the sum of two
penalty terms. Raskutti et al. (2009, 2012)
proposed a doubly penalized estimator, which is the solution of the
minimization of a least-squares criterion penalized by the sum of a
sparsity penalty term and a combination of the -norm with the
Hilbert norm:
where are the tuning parameters that should
be suitably chosen.
Meier et al. (2009) proposed a related family of estimators, based on the
penalization with the empirical -norm. Their penalty function is
the sum of the sparsity penalty term, , and a
smoothness penalty term. Huet and Taupin (2017) considered the same
approximation functional spaces as Durrande et al. (2013), and obtained a
doubly penalized estimator of a meta-model which approximates the
Hoeffding decomposition of . Their estimator is the solution of the
least-squares minimization penalized by the penalty function defined in
Equation ((8)) adapted to the multivariate setting,
This procedure, called RKHS ridge group sparse, estimates the groups
that are suitable for predicting , and the relationship between
and for each group. The obtained estimator, called RKHS
meta-model, is used then to estimate the Sobol indices of . This
approach renders it possible to estimate the Sobol indices for all
groups in the support of the RKHS meta-model, including the interactions
of possibly high order, a point known to be difficult in practice.
In this paper, we introduce an R package, called RKHSMetaMod, that
implements the RKHS ridge group sparse procedure. The functions of this
package allows us to:
implement the RKHS ridge group sparse procedure and a special case
of it, called the RKHS group lasso procedure (when in the
penalty function ((9))), in order to estimate the terms
in the Hoeffding decomposition of the meta-model
leading to an estimation of the function (see Section
Optimization algorithms),
choose the tuning parameters and (see Equation
((9))), using a procedure that leads to obtain the
best RKHS meta-model in terms of the prediction quality,
The current version of the package supports uniformly distributed input
variables on . However, it could be easily adapted
to datasets with input variables from another distribution by making a
small modification to one of its functions (see Remark 3
of Section Calculation of the Gram matrices).
Let us give a brief overview of the related existing statistical
packages to the RKHSMetaMod package. The R package
sensitivity is
designed to implement sensitivity analysis methods and provides the
approaches for numerical calculation of the Sobol indices. In
particular, Kriging method can be used to reduce the number of the
observations in global sensitivity analysis. The function sobolGP of
the package sensitivity builds a Kriging based meta-model using the
function km of the package
DiceKriging(Roustant et al. 2012), and estimates its Sobol indices. This procedure can also
be done using the function km and the function fast99 of the package
sensitivity (see Section 4.5. of Roustant et al. (2012)). In this case, the idea
is once again to build a Kriging based meta-model using the function
km and then estimate its Sobol indices using the function fast99. In
both cases the true function is substituted by a Kriging based
meta-model and then its Sobol indices are estimated. In the sobolGP
function, the Sobol indices are estimated by the Monte Carlo
integration, while the fast99 function estimates them using the
extended-FAST method (Saltelli et al. 1999). To reduce
the computational burden when dealing with large datasets and complex
models, in RKHSMetaMod package, we propose to use the empirical
variances to estimate the Sobol indices (see Section Estimation of the
Sobol indices). Besides, the estimation of the Sobol
indices in the RKHSMetaMod package is done based on the RKHS
meta-model which is a sparse estimator. It is beneficial since instead
of calculating the Sobol indices of all groups , only
the Sobol indices associated with the groups in the support of the RKHS
meta-model are computed (see Section Estimation of the Sobol
indices). Moreover, the functions sobolGP and fast99
provide the estimation of the first order and the total Sobol indices
only, while the procedure in the RKHSMetaMod package makes it possible
to estimate the high order Sobol indices. The R packages DiceKriging
and DiceOptim (Deep
Inside Computer Experiments Kriging/Optim) (Roustant et al. 2012) implement the
Kriging based meta-models to estimate complex models in the high
dimensional context. They provide different GP (Kriging) models
corresponding to the Gaussian, Matérn, Exponential and Power-Exponential
correlation functions. The estimation of the parameters of the
correlation functions in these packages relies on the global optimizer
with gradient genoud algorithm of the package
rgenoud(Mebane and Sekhon 2011).
These packages do not implement any method of the sensitivity analysis
themselves. However, some authors (see Section 4.5. of Roustant et al. (2012) for
example) perform sensitivity analysis on their estimated meta-models by
employing the functions of the package sensitivity. The R package
RobustGaSP (Robust
Gaussian Stochastic Process) (Gu et al. 2019) approximates a complex model by
a GP meta-model. This package implements marginal posterior mode
estimation of the GP model parameters. The estimation method in this
package insures the robustness of the parameter estimation in the GP
model, and allows one also to identify input variables that have no
effect on the variability of the function under study. The R package
mlegp (maximum likelihood
estimates of Gaussian processes) (Dancik and Dorman 2008) provides functions to
implement both meta-modelling approaches and sensitivity analysis
methods. It obtains maximum likelihood estimates of the GP model for the
output of costly computer experiments. The GP models are built either on
the basis of the Gaussian correlation function or on the basis of the
first degree polynomial trend. The sensitivity analysis methods
implemented in this package include Functional Analysis of Variance
(FANOVA) decomposition, plot functions to obtain diagnostic plots, main
effects, and second order interactions. The prediction quality of the
meta-model depends on the quality of the estimation of its parameters
and more precisely the estimation of parameters in the correlation
functions (Kennedy and O’Hagan 2000). The maximum likelihood
estimation of these parameters often produce unstable results, and as a
consequence, the obtained meta-model may have an inferior prediction
quality (Gu et al. 2018; Gu 2019). The RKHSMetaMod
package is devoted to the meta-model estimation on the RKHS
. It implements the convex optimization algorithms to
calculate meta-models; provides the functions to compute the prediction
error of the obtained meta-models; performs the sensitivity analysis on
the obtained meta-models and more precisely calculate their Sobol
indices. The convex optimization algorithms used in this package are all
written using C++ libraries, and are adapted to take into account the
problem of high dimensionality in this context. This package is
available from the Comprehensive R Archive Network (CRAN)
(Kamari 2019).
The organization of the paper is as follows: In the next Section, we
describe the estimation method. In Section Algorithms,
we present in details the algorithms used in the RKHSMetaMod package.
Section RKHSMetaMod through examples includes two
parts: In the first part, Section Simulation study,
the performance of the RKHSMetaMod package functions is validated
through a simulation study. In the second part, Section Comparison
examples, the comparison in terms of the predictive
accuracy between the RKHS meta-model and the Kriging based meta-models
from RobustGaSP(Gu et al. 2019) and DiceKriging(Roustant et al. 2012) packages is
given through two examples.
RKHS ridge group sparse and RKHS group lasso procedures
Let us denote by the number of observations. The dataset consists of
a vector of observations , and a matrix
of features with components
For some tuning
parameters , , , the RKHS ridge group
sparse criterion is defined by,
where represents the matrix of variables corresponding to the
-th group, i.e.
and where is the empirical -norm of
defined by the sample as
The penalty function in the criterion ((10)) is the sum
of the Hilbert norm and the empirical norm, which allows us to select
few terms in the additive decomposition of over sets
. Moreover, the Hilbert norm favours the smoothness
of the estimated , .
Let
be the set of functions. Then the RKHS meta-model is defined by,
According to the Representer Theorem (Kimeldorf and Wahba 1970), the
non-parametric functional minimization problem described above is
equivalent to a parametric minimization problem. Indeed, the solution of
the minimization problem ((11)) belonging to the RKHS
is written as , where for
some matrix
we have for all ,
Let be the Euclidean norm (called also -norm) in
, and for each , let be the
Gram matrix associated with the kernel , i.e.
. Let also be the matrix
that satisfies , and let and
be the minimizers of the following penalized
least-squares criterion:
Then the estimator defined in Equation
((11)) satisfies,
Remark 1. The constraint
is crucial for theoretical properties, but the value of is
generally unknown and has no practical usefulness. In this package, it
is not taken into account in the parametric minimization problem.
For each , let and be the weights
that are chosen suitably. We define and
with .
Remark 2. This formulation simplifies the choice of the tuning
parameters since instead of tuning
parameters and , , only two
parameters and are tuned. Moreover, the weights
and , may be of interest in
practice. For example, one can take weights that increase with the
cardinal of in order to favour the effects with small interaction
order between variables.
For the sake of simplicity, in the rest of this paper for all
the weights and are assumed to be
set as one, and the RKHS ridge group sparse criterion is then expressed
as follows:
If we consider only the second part of the penalty function in the
criterion ((13)) ( i.e. set ), we obtain the
RKHS group lasso criterion as follows:
which is a group lasso criterion (Yuan and Lin 2006)
up to a scale transformation.
In the RKHSMetaMod package, the RKHS ridge group sparse algorithm is
initialized using the solutions obtained by solving the RKHS group lasso
algorithm. Indeed, the penalty function in the RKHS group lasso
criterion ((14)) insures the sparsity in the solution.
Therefore, for a given value of , by implementing the RKHS group
lasso algorithm (see Section RKHS group lasso), a
RKHS meta-model with few terms in its additive decomposition is
obtained. The support and the coefficients of a RKHS meta-model which is
obtained by implementing the RKHS group lasso algorithm will be denoted
by and
, respectively. From now on, we
denote the tuning parameter in the RKHS group lasso criterion by:
Choice of the tuning parameters
While dealing with an optimization problem of a criterion of the form
((13)), one of the essential steps is to choose the
appropriate tuning parameters. Cross-validation is generally used for
that purpose. Nevertheless in the context of high-dimensional complex
models, the computational time for a cross-validation procedure may be
prohibitively high. Therefore, we propose a procedure based on a single
testing dataset:
we first choose, a grid of values of the tuning parameters and
;
Let be the smallest value of (see
Equation ((15))), such that the solution to the
minimization of the RKHS group lasso problem for all
is . We have,
In order to set up the grid of values of , one may find
and then a grid of values of could be
defined by for
The grid of values of is
chosen by the user.
next, for the grid of values of and , we calculate a
sequence of estimators. Each estimator associated with the pair
in the grid of values of and , denoted
by , is the solution of the RKHS ridge
group sparse optimization problem or the RKHS group lasso
optimization problem if .
finally, the obtained estimators are
evaluated using a testing dataset,
.
The prediction error associated with each estimator
is calculated by,
where for being the support of the estimator
we have,
The pair with the smallest value
of the prediction error is chosen, and the estimator
is considered as
the best estimator of the function , in terms of the prediction
error.
In the RKHSMetaMod package, the algorithm to calculate a sequence of
the RKHS meta-models, the value of , and the
prediction error are implemented as RKHSMetMod, mumax, and
PredErr functions, respectively. These functions are described in
Section "Overview of the RKHSMetaMod functions" (supplementary
materials), and illustrated in Example 1, Example 2,
and Examples 1, 2, 3, respectively.
Estimation of the Sobol indices
The variance of the function is estimated by the variance of the
estimator . As the estimator belongs to the
RKHS , it admits the Hoeffding decomposition and,
In order to reduce the computational cost in practice, one may estimate
the variances of , by their
empirical variances. Let be the empirical mean of
, , then:
For the groups that do not belong to the support of ,
we have and for the groups that belong to the
support of , the estimators of the Sobol indices of are
defined by,
In the RKHSMetaMod package, the algorithm to calculate the empirical
Sobol indices , is implemented as
SIemp function. This function is described in Section
"Companion functions" (supplementary materials) and illustrated in
Examples 1, 2, 3.
3 Algorithms
The RKHSMetaMod package implements two optimization algorithms: the
RKHS ridge group sparse (see Algorithm 2) and the RKHS
group lasso (see Algorithm 1). These algorithms rely on the
Gram matrices , that have to be positive
definite. Therefore, the first and essential step in this package is to
calculate these matrices and insure their positive definiteness. The
algorithm of this step is described in the next Section. The second step
is to estimate the RKHS meta-model. In the RKHSMetaMod package, two
different objectives based on different procedures are considered to
calculate this estimator:
The RKHS meta-model with the best prediction quality.
The procedure to calculate the RKHS meta-model with the best
prediction quality has been described in Section Choice of the
tuning parameters: a sequence of values of the tuning
parameters is considered, and the RKHS meta-models
associated with each pair of the values of are
calculated. For , the RKHS meta-model is obtained by
solving the RKHS group lasso optimization problem, while for
the RKHS ridge group sparse optimization problem is
solved to calculate the RKHS meta-model. The obtained estimators are
evaluated by considering a new dataset and the RKHS meta-model with
the minimum value of the prediction error is chosen as the best
estimator.
The RKHS meta-model with at most groups in its support, i.e.
.
First, the tuning parameter is set as zero. Then, a value
of for which the number of groups in the
solution of the RKHS group lasso optimization problem is equal to
, is computed. This value of will be denoted by
. Finally, the RKHS meta-models containing at most
groups in their support are obtained by implementing the RKHS
ridge group sparse algorithm for a grid of values of
and . This procedure is described in more details in
Section RKHS meta-model with active
groups.
Calculation of the Gram matrices
The available kernels in the RKHSMetaMod package are: Gaussian kernel,
Matérn kernel, Brownian kernel, quadratic kernel and linear
kernel. The usual presentation of these kernels is given in Table
1.
Table 1: List of the reproducing kernels used to construct the RKHS
. The range parameters in the Gaussian and Matérn
kernels are assumed to be fixed and set as and
, respectively. The value is added to the Brownian
kernel to relax the constraint of nullity at the origin
(Durrande et al. 2013).
Kernel type
Mathematical formula for
RKHSMetaMod name
Gaussian
"gaussian"
Matérn 3/2
"matern"
Brownian
"brownian"
Quadratic
"quad"
Linear
"linear"
The choice of kernel, that is done by the user, determines the
functional approximation space. For a chosen kernel, the algorithm to
calculate the Gram matrices , in the
RKHSMetaMod package, is implemented as calcKv function. This
algorithm is based on three essential points:
Modify the chosen kernel:
In order to satisfy the conditions of constructing the RKHS
described in Section "RKHS construction" of the
Appendix (supplementary materials), these kernels are modified
according to Equation "(2)" (see the Appendix (supplementary
materials)). Let us take the example of the Brownian kernel:
The RKHS associated with the Brownian kernel
is well known to be
with the inner product
Easy calculations lead to obtain the Brownian kernel as follows,
The RKHS associated with kernel is the set
,
and we have
𝟙
Remark 3. In the current version of the package, we consider
the input variables that are uniformly distributed
on . In order to consider the input variables that are not
distributed uniformly, it suffices to modify a part of the function
calcKv related to the calculation of the kernels ,
. For example, for being distributed
with law on
,
the kernel associated with the Brownian kernel is
calculated as follows,
The other parts of function calcKv remain unchanged.
Calculate the Gram matrices for all :
First, for all , the Gram matrices associated with
kernels are calculated using Equation "(2)" (see the
Appendix (supplementary materials)),
Then, for all
, the Gram matrices associated with kernel
are computed by
.
Insure the positive definiteness of the matrices :
The output of function calcKv is one of the input arguments
of the functions associated with the RKHS group lasso and the RKHS
ridge group sparse algorithms. Throughout these algorithms we need
to calculate the inverse and the square root of the matrices .
In order to avoid the numerical problems and insure the
invertibility of the matrices , it is mandatory to have these
matrices positive definite. One way to render the matrices
positive definite is to add a nugget effect to them. That is, to
modify matrices by adding a diagonal with a constant term,
i.e. . The value of is computed
based on the data and through a part of the algorithm of the
function calckv. Let us briefly explain this part of the
algorithm:
For each group , let be
the eigenvalues associated with the matrix . Set
and
. For some
fixed value of tolerance tol, and for each matrix , if
"tol",
then, the eigenvalues of are replaced by
epsilon, with epsilon being equal to
tol. The value of tol is set as
by default, but one may consider a smaller or a greater
value for it depending on the kernel chosen and the value of .
The function calcKv is described in Section "Companion
functions" (supplementary materials) and illustrated in Example
2.
Optimization algorithms
The RKHS meta-model is the solution of one of the optimization problems:
the minimization of the RKHS group lasso criterion presented in Equation
((14)) (if ), or the minimization of the RKHS ridge
group sparse criterion presented in Equation ((13)) (if
). In the following, the algorithms to solve these
optimization problems are presented.
RKHS group lasso
A popular technique for doing group wise variable selection is group
lasso. With this procedure, depending on the value of the tuning
parameter , an entire group of predictors may drop out of the
model. An efficient algorithm for solving group lasso problem is the
classical block coordinate descent algorithm
(Boyd et al. 2011; Bubeck 2015). Following the idea of
Fu (1998), Yuan and Lin (2006) implemented a
block wise descent algorithm for the group lasso penalized least-squares
under the condition that the model matrices in each group are
orthonormal. A block coordinate (gradient) descent algorithm for solving
the group lasso penalized logistic regression is then developed by
Meier et al. (2008). This algorithm is implemented in the R package
grplasso available from
CRAN (Meier 2020). Yang and Zou (2015) proposed a unified
algorithm named group wise majorization descent for solving the general
group lasso learning problems by assuming that the loss function
satisfies a quadratic majorization condition. The implementation of
their work is done in the
gglasso R package
available from CRAN (Yang et al. 2020).
In order to solve the RKHS group lasso optimization problem, we use the
classical block coordinate descent algorithm. The minimization of
criterion (see Equation ((14))) is done
along each group at a time. At each step of the algorithm, the
criterion is minimized as a function of the current
block’s parameters, while the parameters values for the other blocks are
fixed to their current values. The procedure is repeated until
convergence. This procedure leads to Algorithm 1 (see the
Appendix (supplementary materials) for more details on this procedure).
Algorithm 1: RKHS group lasso algorithm:
In the RKHSMetaMod package, the Algorithm 1 is implemented
as RKHSgrplasso function. This function is described in Section
"Companion functions" (supplementary materials) and illustrated in
Example 2.
RKHS ridge group sparse
In order to solve the RKHS ridge group sparse optimization problem, we
propose an adapted block coordinate descent algorithm. This algorithm is
provided in two steps:
Initialize the input parameters by the
solutions of the RKHS group lasso algorithm for each value of the
tuning parameter , and implement the RKHS ridge group sparse
algorithm through the active support of the RKHS group lasso
solutions until it achieves convergence. This step is provided in
order to decrease the execution time. In fact, instead of
implementing the RKHS ridge group sparse algorithm over the set of
all groups , it is implemented only over the groups in
the support of the solution of the RKHS group lasso algorithm,
.
Re-initialize the input parameters with the
obtained solutions of Step and implement the RKHS
ridge group sparse algorithm through all groups in
until it achieves convergence. This second step makes it possible to
verify that no group is missing in the output of Step .
This procedure leads to Algorithm 2 (see the Appendix
(supplementary materials) for more details on this procedure).
Algorithm 2: RKHS ridge group sparse algorithm:
In the RKHSMetaMod package the Algorithm 2 is implemented
as penMetMod function. This function is described in Section
"Companion functions" (supplementary materials) and illustrated in
Example 2.
RKHS meta-model with at most groups in its support
By considering some prior information about the data, one may be
interested in a RKHS meta-model with the number of groups
in its support not greater than some "". In order to obtain such
an estimator, we provide the following procedure in the RKHSMetaMod
package:
First, the tuning parameter is set as zero and a value of
for which the solution of the RKHS group lasso algorithm,
Algorithm 1, contains exactly groups in its
support is computed. This value is denoted by .
Then, the RKHS ridge group sparse algorithm, Algorithm
2, is implemented by setting the tuning parameter
equal to and a grid of values of the tuning parameter
.
Algorithm 3: Algorithm to estimate RKHS meta-model with at
most qmax
groups in its support:
This algorithm is implemented in the RKHSMetaMod package, as function
RKHSMetModqmax (see Section "Main RKHSMetaMod functions"
(supplementary materials) for more details on this function).
Remark 4. As both terms in the penalty function of criterion
((13)) enforce sparsity to the solution, the estimator
obtained by solving the RKHS ridge group sparse associated with the pair
of the tuning parameters may contain a smaller
number of groups than the solution of the RKHS group lasso optimization
problem (i.e. the RKHS ridge group sparse with ).
And therefore, the estimated RKHS meta-model contains at most ""
groups in its support.
4 RKHSMetaMod through examples
Simulation study
Let us consider the g-function of Sobol (Saltelli et al. 2009) in
the Gaussian regression framework, i.e. . The error
term is a centered Gaussian random variable with variance
, and is the g-function of Sobol defined over
by,
The Sobol indices of the g-function can be expressed analytically:
Set , , and . With these
values of coefficients , the variables and explain
of the variance of function (see Table 4).
In this Section, three examples are presented. In all examples, the
value of Dmax is set as three. Example 1 illustrates the use of
the RKHSMetMod function by considering three different kernels,
"matern", "brownian", and "gaussian" (see Table
1), and three datasets of observations
and input variables. The larger datasets with
observations and input variables are
studied in Examples 2 and 3. In each example, two
independent datasets are generated: to estimate the meta-models,
and to estimate the prediction errors. The design matrices
and are the Latin Hypercube Samples of the input variables that are
generated using maximinLHS function of the package
lhs available at CRAN
(Carnell 2021):
library(lhs); X <-maximinLHS(n, d); XT <-maximinLHS(n, d)
The response variables and are calculated as
and , where ,
and are centered Gaussian random variables with
:
a <-c(0.2, 0.6, 0.8, 100, 100, 100, 100, 100, 100, 100)[1:d]g=1; for (i in1:d) g = g*(abs(4*X[,i]-2)+a[i])/(1+a[i])sigma <-0.2epsilon <-rnorm(n, 0, sigma^2); Y <- g + epsilongT=1; for (i in1:d) gT = gT*(abs(4*XT[,i]-2)+a[i])/(1+a[i])epsilonT <-rnorm(n, 0, sigma^2); YT <- gT + epsilonT
Example 1. RKHS meta-model estimation using RKHSMetMod function:
In this example, three datasets of points maximinLHS over
with and are generated, and a grid
of five values of tuning parameters and is considered as
follows:
For each dataset, the experiment is repeated times. At each
repetition, the RKHS meta-models associated with the pair of tuning
parameters are estimated using the RKHSMetMod function:
These meta-models are evaluated using a testing dataset. The prediction
errors are computed for them using the PredErr function. The RKHS
meta-model with minimum prediction error is chosen to be the best
estimator for the model. Finally, the Sobol indices are computed for the
best RKHS meta-model using the function SIemp:
The vector mu is the values of the tuning parameter that are
calculated throughout the function RKHSMetMod. It could be recovered
from the output of the RKHSMetMod function as follows:
mu <-vector()l <-length(gamma); for(i in1:length(frc)){mu[i] <- res[[(i-1)*l+1]]$mu}
The performance of this method for estimating a meta-model is evaluated
by considering a third dataset ,
, with . The global prediction error is calculated as
follows:
Let be the best RKHS meta-model obtained in the
repetition , , then
The values of obtained for different kernels and values of are
given in Table 2.
Table 2: Example 1: The columns of the table correspond to
the different datasets with and . Each line
of the table, from up to down, gives the value of GPE obtained for
each dataset associated with the "matern", "brownian" and
"gaussian" kernels, respectively.
As expected the value of decreases as increases. The lowest
values of are obtained when using the "matern" kernel.
In order to sum up the behaviour of the procedure for estimating the
Sobol indices, we consider the mean square error (MSE) criterion
obtained by ,
where for each group , denotes the true values of the Sobol
indices, and is the empirical Sobol indices of the
best RKHS meta-model in repetition . The obtained values of MSE for
different kernels and values of are given in Table
3.
Table 3: Example 1: The columns of the table correspond to
the different datasets with and . Each line
of the table, from up to down, gives the value of MSE obtained for
each dataset associated with the "matern", "brownian" and
"gaussian" kernels, respectively.
As expected, the values of MSE are smaller for larger values of . The
smallest values are obtained when using the "matern" kernel.
The means of the empirical Sobol indices of the best RKHS meta-models
through all repetitions for and "matern" kernel are
displayed in Table 4.
Table 4: Example 1: The first line of the table gives the
true values of the Sobol indices . The second line gives
the mean of the estimated empirical Sobol indices
()
greater than calculated over fifty simulations
for and "matern" kernel. The sum of the Sobol indices is
displayed in the last column.
sum
43.30
24.30
19.20
5.63
4.45
2.50
0.57
99.98
46.10
26.33
20.62
2.99
2.22
1.13
0.0
99.39
It appears that the estimated Sobol indices are close to the true ones,
nevertheless, they are overestimated for the main effects, i.e. groups
, and underestimated for the interactions of
order two and three, i.e. groups
. Note that the strategy of
choosing the tuning parameters is based on the minimization of the
prediction error of the estimated meta-model, which may not minimize the
error of estimating the Sobol indices.
Taking into account the results obtained for this Example 1,
the calculations in the rest of the examples is done using only the
"matern" kernel.
Example 2. A time-efficient strategy to obtain the "optimal"
tuning parameters when dealing with large datasets:
A dataset of points maximinLHS over with and
is generated. First, we use functions calcKv and
mumax to compute the eigenvalues and eigenvectors of the
positive definite matrices , and the value of ,
respectively:
Set and,
Calculate the
RKHS meta-models associated with the values of
by using the function RKHSgrplasso.
Gather the obtained RKHS meta-models in a list, resg (while
this job could be done using the function RKHSMetMod by setting
, in this example, we use the function RKHSgrplasso in
order to avoid the re-calculation of ’s at the next step).
Thereafter, for each estimator in resg, the prediction error
is calculated by considering a new dataset and using the function
PredErr. The value of with the smallest error of prediction
in this step is denoted by . Let us implement this step:
For a grid of values of , a sequence of the RKHS meta-models
are calculated and gathered in the resg list:
mu_g <-c(mumax*0.5^(2:10)) res_g <-list(); resg <-list()for(i in1:length(mu_g)){ resg[[i]] <-RKHSgrplasso(Y, Kv, mu_g[i], 1000, FALSE) res_g[[i]] <-list("mu_g"=mu_g, "gamma"=0, "MetaModel"=resg[[i]]) }```Output `res`$\_$`g` contains nine RKHS meta-models and they areevaluated using a testing dataset:``` rgamma <-c(0); Err_g <-PredErr(X, XT, YT, mu_g, gamma, res_g, kernel, Dmax)
The prediction errors of the RKHS meta-models obtained in this step
are displayed in Table 5.
Table 5: Example 2: Prediction errors associated with
the RKHS meta-models computed in step 1.
It appears that the minimum prediction error corresponds to the
solution of the RKHS group lasso algorithm with , so
.
Choose a smaller grid of values of ,
, and set a grid of values of
. Calculate the RKHS meta-models associated with each pair
of the tuning parameters by the function
penMetMod. Calculate the prediction errors for the new
sequence of the RKHS meta-models using the function PredErr.
Compute the empirical Sobol indices for the best estimator. Let us
go back to the implementation of the example and apply this step
:
The grid of the values of in this step is
The RKHS meta-models associated with
this grid of the values of are gathered in a new list
resgnew. We set , and we
calculate the RKHS meta-models for this new grid of the values of
using penMetMod function:
The output res is a list of twelve RKHS meta-models. These
meta-models are evaluated using a new dataset, and their prediction
errors are displayed in Table 6.
Table 6: Example 2: Obtained prediction errors in step 2.
The minimum prediction error is associated with the pair
, and the best RKHS meta-model is then
.
The performance of this procedure for estimating the Sobol indices
is evaluated using the relative error (RE) defined as follows:
For each , let be the true value of the Sobol indices
displayed in Table 4 and be the estimated
empirical Sobol indices. Then
In Table 7 the estimated empirical Sobol indices,
their sum, and the value of RE are displayed.
Table 7: Example 2: The estimated empirical Sobol
indices greater than . The last two columns
show and RE, respectively.
sum
RE
The obtained RE for each group is smaller than , therefore,
the estimated Sobol indices in this example are very close to the true
values of the Sobol indices displayed in the first row of Table
4.
Example 3. Dealing with larger datasets:
Two datasets of points maximinLHS over with
and are generated. In order to obtain one
RKHS meta-model associated with one pair of the tuning parameters
, the number of coefficients to be estimated is equal to
vMax. Table 8 gives the execution time
for different functions used throughout the Examples
1-3. In all examples we used a cluster of computers
with: 2 Intel Xeon E5-2690 processors (2.90GHz) and 96Gb Ram (6x16Gb of
memory 1600MHz).
Table 8: Example 3: The kernel used is "matern". The
execution time for the functions RKHSgrplasso and penMetMod
is displayed in each row for two pairs of values of the tuning
parameters on up,
and on below. In
the column , the number of groups in the
support of each estimated RKHS meta-model is displayed.
calcKv
mumax
RKHSgrplasso
penMetMod
sum
(100,5)
0.09s
0.01s
1s
2s
18
3s
2s
3s
19
5s
(500,10)
33s
9s
247s
333s
39
10min
599s
816s
64
24min
(1000,10)
197s
53s
959s
1336s
24
42min
2757s
4345s
69
2h
(2000,10)
1498s
420s
3984s
4664s
12
2h:56min
12951s
22385s
30
10h:20min
(5000,10)
34282s
6684s
38957s
49987s
11
36h:05min
99221s
111376s
15
69h:52min
As we can see, the execution time increases fast as increases. In
Figure 1 the plot of the logarithm of the time (in
seconds) versus the logarithm of is displayed for the functions
calcKv, mumax, RKHSgrplasso and penMetMod.
Figure 1: Example 3: Timing plot for ,
, and different functions of the
RKHSMetaMod package. The logarithm of the execution time (in seconds)
for the functions RKHSgrplasso and penMetMod is displayed for
two pairs of values of the tuning parameters
in solid lines, and
in dashed
lines.
It appears that, the algorithms of these functions are of polynomial
time with for the functions
calcKv and mumax, and for the
functions RKHSgrplasso and penMetMod.
Taking into account the results obtained for the prediction error and
the values of in Example 2,
in this example, only two values of the tuning parameter
, and one value of the
tuning parameter are considered. The RKHS meta-models
associated with the pair of values , are
estimated using the RKHSMetMod function:
Table 9 gives the estimated empirical Sobol indices as
well as their sum, the values of RE (see Equation ((18))),
and the prediction errors associated with the obtained estimators.
Table 9: Example 3: The estimated empirical Sobol indices
greater than associated with each estimated RKHS
meta-model is printed. The last three columns show
, RE, and the prediction error (Err),
respectively. We have ,
and .
sum
RE
Err
For we obtained the smaller values of RE and prediction error
(Err). So as expected, the estimation of the Sobol indices as well as
the prediction quality are better for larger values of .
In Figure 2 the result of the estimation quality and the Sobol
indices for the dataset with equal to , equal to , and
are displayed.
Figure 2: Example 3: On the left, the RKHS meta-model versus
the g-function is plotted. On the right, the empirical Sobol indices in
the y-axis and vMax groups in the x-axis are
displayed.
The line in red crosses the cloud of points as long as the
values of the g-function are smaller than three. When the values of the
g-function are greater than three, the estimator tends to
underestimate the g-function. Concerning the Sobol indices obtained by
the estimator , as illustrated in the right-hand plot, with
the exception of groups , , , , ,
and for which we obtained significant values of the sobol
indices, for all other groups the estimated sobol indices are very small
and almost zero.
Comparison examples
This section includes two examples. In the first example we reproduce an
example from paper Gu et al. (2019) and compare the prediction quality of the
RKHS meta-model with the GP (Kriging) based meta-models from the
RobusGaSP(Gu et al. 2019) and DiceKriging(Roustant et al. 2012) packages. The
objective is to evaluate the quality of the RKHS meta-model and to
compare it with methods recently proposed for approximating complex
models. In the first example we consider one-dimensional model and focus
on the comparison between the true model and the estimated meta-model.
In the second example we reproduce an example from paper Roustant et al. (2012)
which allows us to compare the prediction quality of the RKHS meta-model
with the Kriging based meta-model from DiceKriging package, as well as
the estimation quality of the Sobol indices in our package with the
well-known package sensitivity. For the sake of comparison between the
three methods, the meta-models are calculated using the same
experimental design and outputs, and the same kernel function available
in three packages is used. However, in packages RobustGaSP and
DiceKriging the range parameter (see Table 1) in the
kernel function is estimated by marginal posterior modes with the robust
parameterization and by MLE with upper and lower bounds, respectively,
while it is assumed to be fixed and set as in the
RKHSMetaMod package.
Example 4. "The modified sine wave function":
We consider the -dimensional modified sine wave function defined by
over . The same experimental
design as described in Gu et al. (2019) is considered: the design matrix is
a sequence of equally spaced points on , and the response
variable is calculated as :
X <-as.matrix(seq(0,1,1/11)); Y <-sinewave(X)
where sinewave function is defined in Gu et al. (2019). We build the GP based
meta-models by the RobustGaSP and the DiceKriging packages using the
constant mean function and kernel Matérn :
As , we have . We consider the grid of values of
and
. The RKHS meta-models associated
with the pair of values , ,
are estimated using the RKHSMetMod function:
Given a testing dataset , the prediction errors associated with
the obtained RKHS meta-models are calculated using the PredErr
function, and the best RKHS meta-model is chosen to be the estimator
of the model :
To compare these three estimators in terms of the prediction quality, we
perform prediction on test points, equally spaced in :
predict_X <-as.matrix(seq(0,1,1/99))#prediction with the GP based meta-models:rgasp.predict <-predict(res.rgasp, predict_X)km.predict <-predict(res.km, predict_X, type='UK')#prediction with the best RKHS meta-model:res.predict <-prediction(X, predict_X, kernel, Dmax, res, Err)
The prediction results are plotted in Figure 3. The black
circles that correspond to the prediction from the RKHSMetMod package
are closer to the real output than the green and the blue circles
corresponding to the predictive means from the RobustGaSP and
DiceKriging packages.
Figure 3: Example 4: Prediction of the
modified sine wave function with test points, equally spaced in
. The x-axis is the real output and the y-axis is the prediction.
The black circles are the prediction from RKHSMetMod, the green
circles are the predictive mean from RobustGaSP, and the blue circles
are the predictive mean from DiceKriging.
The meta-model results are plotted in Figure 4. The
prediction from the RKHSMetaMod package plotted as the black curve is
much more accurate as an estimate of the true function (plotted in red)
than the predictive mean from the RobustGaSP and DiceKriging
packages plotted as the blue and green curves, respectively. As already
noted by Gu et al. (2019), for that sine wave example, the meta-model from the
DiceKriging package "degenerates to the fitted mean with spikes at the
design points".
Figure 4: Example 4: The red curve is the
graph of the modified sine wave function with test points, equally
spaced in . The black curve is the prediction produced by the
RKHSMetaMod package. The blue curve is the predictive mean by the
DiceKriging package, and the green curve is the predictive mean
produced by the RobustGaSP package.
Example 5. "A standard SA 8-dimensional example":
We consider the -dimensional g-function of Sobol implemented in the
package sensitivity: the function as defined in Equation
((17)) with coefficients
. With these
values of coefficients , the variables , , and
explain of the variance of function (see Table
10).
We consider the same experimental design as described in Roustant et al. (2012):
the design matrices and are -point optimal Latin Hypercube
Samples of the input variables generated by the optimumLHS function of
package lhs, and the response variables and are calculated as
, and using sobol.fun function of the package
sensitivity:
n <-80; d <-8library(lhs); X <-optimumLHS(n, d); XT <-optimumLHS(n, d)library(sensitivity); Y <-sobol.fun(X); YT <-sobol.fun(XT)
Let us first consider the RKHS meta-model method. We set Dmax, and
we consider the grid of values of
, and
. The RKHS meta-models associated
with the pair of values , ,
are estimated using the RKHSMetMod function:
Given the testing dataset , the prediction errors associated
with the obtained RKHS meta-models are calculated using PredErr
function, and the best RKHS meta-model is chosen to be the estimator
of the model . Finally, the Sobol indices are computed for the
best RKHS meta-model using the function SIemp:
Secondly, let us build the GP based meta-model. We use the km function
of the package DiceKriging with the constant mean function and kernel
Matérn :
library(DiceKriging)res.km <-km(design = X, response = Y, covtype ="matern3_2")
The Sobol indices associated with the estimated GP based meta-model are
calculated using fast99 function of the package sensitivity:
SI.km <-fast99(model = kriging.mean, factors = d, n =1000,q ="qunif", q.arg =list(min =0, max =1), m = res.km)
where kriging.mean function is defined in Roustant et al. (2012).
The result of the estimation with the best RKHS meta-model and the
Kriging based meta-model is drawn in Figure 5.
Figure 5: Example 5: The x-axis is the real output
and the y-axis is the fitted meta-model. The black circles are the
meta-model from RKHSMetMod and the blue circles are the meta-model
from DiceKriging.
The black circles that correspond to the best RKHS meta-model are
closer to the real output than the blue circles corresponding to the GP
based meta-model from the DiceKriging package. Another way to evaluate
the prediction quality of the estimated meta-models is to consider the
mean square error of the fitted meta-model computed by
. We obtained
and for the Kriging based meta-model and the RKHS meta-model,
respectively, which confirms the good behavior of the RKHS meta-model.
The estimated Sobol indices associated with the RKHS meta-model and the
Kriging based meta-model are given in Table 10.
Table 10: Example 5: The true values of the Sobol
indices greater than are given in the first raw.
The estimated Sobol indices associated with the RKHS meta-model
() and the Kriging based meta-model
() are given in second and third rows,
respectively.
sum
71.62
17.90
2.37
0.72
5.97
0.79
0.24
0.20
0.06
0.07
0.02
99.96
75.78
17.42
1.71
0.47
4.00
0.05
0.07
0.28
0.09
0.00
0.00
99.87
71.18
15.16
1.42
0.44
0.00
0.00
0.00
0.00
0.00
0.00
0.00
88.20
As shown, with RKHS meta-model, we obtained non-zero values for the
interactions of order two. Concerning the main effects, excepting the
first one, the estimated Sobol indices with the RKHS meta-model are
closer to the true ones. However, the interactions of order three are
ignored by both meta-models. For a general comparison of the estimation
quality of the Sobol indices, one may consider the criterion RE defined
in Equation ((18)), which is equal to for the Kriging
based meta-model, and for the RKHS meta-model. Comparing the
values of RE, we can point out that the Sobol indices are better
estimated with the RKHS meta-model in that model.
5 Summary and discussion
In this paper, we proposed an R package, called RKHSMetaMod, that
estimates a meta-model of a complex model . This meta-model belongs
to a reproducing kernel Hilbert space constructed as a direct sum of
Hilbert spaces (Durrande et al. 2013). The estimation of the meta-model is
carried out via a penalized least-squares minimization allowing both to
select and estimate the terms in the Hoeffding decomposition, and
therefore, to select the Sobol indices that are non-zero and estimate
them (Huet and Taupin 2017). This procedure makes it possible to estimate
the Sobol indices of high order, a point known to be difficult in
practice. Using the convex optimization tools, RKHSMetaMod package
implements two optimization algorithms: the minimization of the RKHS
ridge group sparse criterion ((13)) and the RKHS group
lasso criterion ((14)). Both of these algorithms rely on the
Gram matrices , and their positive definiteness.
Currently, the package considers only uniformly distributed input
variables. If one is interested by another distribution of the input
variables, it suffices to modify the calculation of the kernels
, in the function calcKv of this package
(see Remark 3). The available kernels in the
RKHSMetaMod package are: Gaussian kernel (with the fixed range
parameter ), Matérn kernel (with the fixed range parameter
), Brownian kernel, quadratic kernel and linear kernel
(see Table 1). With regard to the problem being under study,
one may consider other kernels or kernels with different values of the
range parameter and add them easily to the list of the kernels in
the calcKv function. For the large values of and the
calculation and storage of eigenvalues and eigenvectors of all the Gram
matrices , require a lot of time and a very large
amount of memory. In order to optimize the execution time and also the
storage memory, except for a function that is written in R, all of the
functions of RKHSMetaMod package are written using the efficient C++
libraries through RcppEigen and RcppGSL packages. These functions
are then interfaced with the R environment in order to contribute a user
friendly package.
This article is converted from a Legacy LaTeX article using the
texor package.
The pdf version is the official version. To report a problem with the html,
refer to CONTRIBUTE on the R Journal homepage.
Footnotes
References
G. Blatman and B. Sudret. Adaptive sparse polynomial chaos expansion based on least angle regression. Journal of computational Physics, 230: 2345–2367, 2011.
S. Boyd, N. Parikh, E. Chu, B. Peleato and J. Eckstein. Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers. Found. Trends Mach. Learn., 3(1): 1–122, 2011. URL https://doi.org/10.1561/2200000016.
S. Bubeck. Convex Optimization: Algorithms and Complexity. Found. Trends Mach. Learn., 8(3-4): 231–357, 2015. URL https://doi.org/10.1561/2200000050.
G. M. Dancik and K. S. Dorman. Mlegp: Statistical analysis for computer models of biological systems using R. Bioinformatics, 24(17): 1966, 2008.
N. Durrande, D. Ginsbourger and O. Roustant. Additive Covariance kernels for high-dimensional Gaussian Process modeling. Annales de la faculté des sciences de Toulouse Mathématiques, 21(3): 481–499, 2012. URL http://eudml.org/doc/251000.
N. Durrande, D. Ginsbourger, O. Roustant and L. Carraro. ANOVA kernels and RKHS of zero mean functions for model-based sensitivity analysis. Journal of Multivariate Analysis, 115: 57–67, 2013. URL http://www.sciencedirect.com/science/article/pii/S0047259X1200214X.
W. J. Fu. Penalized Regressions: The Bridge versus the Lasso. Journal of Computational and Graphical Statistics, 7(3): 397–416, 1998. URL http://www.jstor.org/stable/1390712.
M. Gu. Jointly Robust Prior for Gaussian Stochastic Process in Emulation, Calibration and Variable Selection. Bayesian Analysis, 14(3): 857–885, 2019. URL https://doi.org/10.1214/18-BA1133.
M. Gu, J. Palomo and O. Berger James. RobustGaSP: Robust Gaussian Stochastic Process Emulation in R. The R Journal, 11(1): 112, 2019. URL http://dx.doi.org/10.32614/rj-2019-011.
M. Gu, X. Wang and J. O. Berger. Robust Gaussian stochastic process emulation. The Annals of Statistics, 46(6A): 3038–3066, 2018. URL https://doi.org/10.1214/17-AOS1648.
T. Hastie, R. Tibshirani and M. Wainwright. Statistical learning with sparsity: The lasso and generalizations. Chapman & Hall/CRC, 2015.
H. Kamari. RKHSMetaMod: Ridge group sparse optimization problem for estimation of a meta model based on reproducing kernel hilbert spaces. 2019. URL https://CRAN.R-project.org/package=RKHSMetaMod. R package version 1.1.
M. C. Kennedy and A. O’Hagan. Bayesian Calibration of Computer Models. Journal of the Royal Statistical Society, Series B, Methodological, 63: 425–464, 2000.
G. S. Kimeldorf and G. Wahba. A Correspondence Between Bayesian Estimation on Stochastic Processes and Smoothing by Splines. Ann. Math. Statist., 41(2): 495–502, 1970. URL http://dx.doi.org/10.1214/aoms/1177697089.
J. P. C. Kleijnen. Design and analysis of simulation experiments. 1st ed Springer Publishing Company, Incorporated, 2007.
L. Le Gratiet, S. Marelli and B. Sudret. Metamodel-based sensitivity analysis: Polynomial chaos expansions and gaussian processes. In Handbook of uncertainty quantification, Eds R. Ghanem, D. Higdon and H. Owhadi pages. 1289–1325 2017. Cham: Springer International Publishing. ISBN 978-3-319-12385-1. URL https://doi.org/10.1007/978-3-319-12385-1_38.
Loic. Le Gratiet, Claire. Cannamela and Bertrand. Iooss. A Bayesian Approach for Global Sensitivity Analysis of (Multifidelity) Computer Codes. SIAM/ASA Journal on Uncertainty Quantification, 2(1): 336–363, 2014. URL
https://doi.org/10.1137/130926869
.
Y. Lin and H. H. Zhang. Component selection and smoothing in multivariate nonparametric regression. Ann. Statist., 34(5): 2272–2297, 2006. URL https://doi.org/10.1214/009053606000000722.
WR. Mebane and JS. Sekhon. Genetic Optimization Using Derivatives: The rgenoud Package for R. Journal of Statistical Software, Articles, 42(11): 1–26, 2011. URL https://www.jstatsoft.org/v042/i11.
L. Meier, S. van de Geer and P. Bühlmann. High-dimensional additive modeling. Ann. Statist., 37(6B): 3779–3821, 2009. URL https://doi.org/10.1214/09-AOS692.
L. Meier, S. van de Geer and P. Bühlmann. The group lasso for logistic regression. Journal of the Royal Statistical Society. Series B, 70(1): 53–71, 2008. DOI 10.1111/j.1467-9868.2007.00627.x.
G. Raskutti, M. J. Wainwright and B. Yu. Lower bounds on minimax rates for nonparametric regression with additive sparsity and smoothness. In Advances in neural information processing systems, 2009.
G. Raskutti, M. J. Wainwright and B. Yu. Minimax-optimal Rates for Sparse Additive Models over Kernel Classes via Convex Programming. J. Mach. Learn. Res., 13(1): 389–427, 2012. URL http://dl.acm.org/citation.cfm?id=2503308.2188398.
O. Roustant, D. Ginsbourger and Y. Deville. DiceKriging, DiceOptim: Two R Packages for the Analysis of Computer Experiments by Kriging-Based Metamodeling and Optimization. Journal of Statistical Software, Articles, 51(1): 1–55, 2012. URL https://www.jstatsoft.org/v051/i01.
I. M. Sobol. Sensitivity Estimates for Nonlinear Mathematical Models. In Sensitivity estimates for nonlinear mathematical models, 1993.
C. Soize and R. Ghanem. Physical Systems with Random Uncertainties: Chaos Representations with Arbitrary Probability Measure. SIAM Journal on Scientific Computing, 26(2): 395–410, 2004. URL
https://doi.org/10.1137/S1064827503424505
.
A. W. Van der Vaart. Asymptotic statistics. Cambridge University Press, 1998. DOI 10.1017/CBO9780511802256.
W. J. Welch, Robert. J. Buck, J. Sacks, H. P. Wynn, T. J. Mitchell and M. D. Morris. Screening, Predicting, and Computer Experiments. Technometrics, 34(1): 15–25, 1992. URL http://www.jstor.org/stable/1269548.
Y. Yang and H. Zou. A Fast Unified Algorithm for Solving Group-lasso Penalize Learning Problems. Statistics and Computing, 25(6): 1129–1141, 2015. URL http://dx.doi.org/10.1007/s11222-014-9498-5.
Y. Yang, H. Zou and S. Bhatnagar. Gglasso: Group lasso penalized learning using a unified BMD algorithm. 2020. URL https://CRAN.R-project.org/package=gglasso. R package version 1.5.
Text and figures are licensed under Creative Commons Attribution CC BY 4.0. The figures that have been reused from other sources don't fall under this license and can be recognized by a note in their caption: "Figure from ...".
Citation
For attribution, please cite this work as
Kamari, et al., "RKHSMetaMod: An R Package to Estimate the Hoeffding Decomposition of a Complex Model by Solving RKHS Ridge Group Sparse Optimization Problem", The R Journal, 2022
BibTeX citation
@article{RJ-2022-003,
author = {Kamari, Halaleh and Huet, Sylvie and Taupin, Marie-Luce},
title = {RKHSMetaMod: An R Package to Estimate the Hoeffding Decomposition of a Complex Model by Solving RKHS Ridge Group Sparse Optimization Problem},
journal = {The R Journal},
year = {2022},
note = {https://rjournal.github.io/},
volume = {14},
issue = {1},
issn = {2073-4859},
pages = {101-122}
}