Help for package RMSS

Type:

Package

Title:

Robust Multi-Model Subset Selection

Version:

1.2.0

Date:

2025-07-17

Maintainer:

Anthony Christidis <anthony.christidis@stat.ubc.ca>

Description:

Efficient algorithms for generating ensembles of robust, sparse and diverse models via robust multi-model subset selection (RMSS). The robust ensembles are generated by minimizing the sum of the least trimmed square loss of the models in the ensembles under constraints for the size of the models and the sharing of the predictors. Tuning parameters for the robustness, sparsity and diversity of the robust ensemble are selected by cross-validation.

License:

GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]

Encoding:

UTF-8

Biarch:

true

Imports:

Rcpp (≥ 1.0.9), srlars, robStepSplitReg, cellWise, robustbase

Suggests:

testthat, mvnfast

LinkingTo:

Rcpp, RcppArmadillo

RoxygenNote:

7.2.3

NeedsCompilation:

yes

Packaged:

2025-07-17 15:16:43 UTC; anthony

Author:

Anthony Christidis [aut, cre], Gabriela Cohen-Freue [aut]

Repository:

CRAN

Date/Publication:

2025-07-17 15:40:13 UTC

Robust Multi-Model Subset Selection

Description

RMSS performs robust multi-model subset selection.

Usage

RMSS(
  x,
  y,
  n_models,
  h_grid,
  t_grid,
  u_grid,
  initial_estimator = c("robStepSplitReg", "srlars")[1],
  tolerance = 0.1,
  max_iter = 1000,
  neighborhood_search = FALSE,
  neighborhood_search_tolerance = 0.1
)

Arguments

x

Design matrix.

y

Response vector.

n_models

Number of models into which the variables are split.

h_grid

Grid for robustness parameter.

t_grid

Grid for sparsity parameter.

u_grid

Grid for diversity parameter.

initial_estimator

Method used for initial estimator. Must be one of "robStepSplitReg" (default) or "srlars".

tolerance

Tolerance level for convergence of PSBGD algorithm.

max_iter

Maximum number of iterations in PSBGD algorithm.

neighborhood_search

Neighborhood search to improve solution. Default is FALSE.

neighborhood_search_tolerance

Tolerance parameter for neighborhood search. Default is 1e-1.

Value

An object of class RMSS

Author(s)

Anthony-Alexander Christidis, anthony.christidis@stat.ubc.ca

Examples

# Simulation parameters
n <- 50
p <- 100
rho <- 0.8
rho.inactive <- 0.2
group.size <- 5
p.active <- 15
snr <- 2
contamination.prop <- 0.3

# Setting the seed
set.seed(0)

# Block Correlation
sigma.mat <- matrix(0, p, p)
sigma.mat[1:p.active, 1:p.active] <- rho.inactive
for(group in 0:(p.active/group.size - 1))
  sigma.mat[(group*group.size+1):(group*group.size+group.size),
            (group*group.size+1):(group*group.size+group.size)] <- rho
diag(sigma.mat) <- 1

# Simulation of beta vector
true.beta <- c(runif(p.active, 0, 5)*(-1)^rbinom(p.active, 1, 0.7), 
               rep(0, p - p.active))

# Setting the SD of the variance
sigma <- as.numeric(sqrt(t(true.beta) %*% sigma.mat %*% true.beta)/sqrt(snr))

# Simulation of test data
m <- 2e3
x_test <- mvnfast::rmvn(m, mu = rep(0, p), sigma = sigma.mat)
y_test <- x_test %*% true.beta + rnorm(m, 0, sigma)

# Simulation of uncontaminated data 
x <- mvnfast::rmvn(n, mu = rep(0, p), sigma = sigma.mat)
y <- x %*% true.beta + rnorm(n, 0, sigma)

# Contamination of data 
contamination_indices <- 1:floor(n*contamination.prop)
k_lev <- 2
k_slo <- 100
x_train <- x
y_train <- y
beta_cont <- true.beta
beta_cont[true.beta!=0] <- beta_cont[true.beta!=0]*(1 + k_slo)
beta_cont[true.beta==0] <- k_slo*max(abs(true.beta))
for(cont_id in contamination_indices){
 
 a <- runif(p, min = -1, max = 1)
 a <- a - as.numeric((1/p)*t(a) %*% rep(1, p))
  x_train[cont_id,] <- mvnfast::rmvn(1, rep(0, p), 0.1^2*diag(p)) + k_lev * a / 
                        as.numeric(sqrt(t(a) %*% solve(sigma.mat) %*% a))
  y_train[cont_id] <- t(x_train[cont_id,]) %*% beta_cont
}

# RMSS
rmss_fit <- RMSS(x = x_train, y = y_train,
                 n_models = 3,
                 h_grid = c(35), t_grid = c(6, 8, 10), u_grid = c(1:3),
                 initial_estimator = "robStepSplitReg",
                 tolerance = 1e-1,
                 max_iter = 1e3,
                 neighborhood_search = FALSE,
                 neighborhood_search_tolerance = 1e-1)
rmss_coefs <- coef(rmss_fit, 
                   h_ind = 1, t_ind = 2, u_ind = 1,
                   group_index = 1:rmss_fit$n_models)
sens_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/p.active
spec_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/sum(rmss_coefs[-1]!=0)
rmss_preds <- predict(rmss_fit, newx = x_test,
                      h_ind = 1, t_ind = 2, u_ind = 1,
                      group_index = 1:rmss_fit$n_models,
                      dynamic = FALSE)
rmss_mspe <- mean((y_test - rmss_preds)^2)/sigma^2

Coefficients for RMSS Object

Description

coef.RMSS returns the coefficients for a RMSS object.

Usage

## S3 method for class 'RMSS'
coef(
  object,
  h_ind,
  t_ind,
  u_ind,
  individual_models = FALSE,
  group_index = NULL,
  ...
)

Arguments

object

An object of class RMSS.

h_ind

Index for robustness parameter.

t_ind

Index for sparsity parameter.

u_ind

Index for diversity parameter.

individual_models

Argument to determine whether the coefficients of each model are returned. Default is FALSE.

group_index

Groups included in the ensemble. Default setting includes all the groups.

...

Additional arguments for compatibility.

Value

The coefficients for the RMSS object.

Author(s)

Anthony-Alexander Christidis, anthony.christidis@stat.ubc.ca

Examples

# Simulation parameters
n <- 50
p <- 100
rho <- 0.8
rho.inactive <- 0.2
group.size <- 5
p.active <- 15
snr <- 2
contamination.prop <- 0.3

# Setting the seed
set.seed(0)

# Block Correlation
sigma.mat <- matrix(0, p, p)
sigma.mat[1:p.active, 1:p.active] <- rho.inactive
for(group in 0:(p.active/group.size - 1))
  sigma.mat[(group*group.size+1):(group*group.size+group.size),
            (group*group.size+1):(group*group.size+group.size)] <- rho
diag(sigma.mat) <- 1

# Simulation of beta vector
true.beta <- c(runif(p.active, 0, 5)*(-1)^rbinom(p.active, 1, 0.7), 
               rep(0, p - p.active))

# Setting the SD of the variance
sigma <- as.numeric(sqrt(t(true.beta) %*% sigma.mat %*% true.beta)/sqrt(snr))

# Simulation of test data
m <- 2e3
x_test <- mvnfast::rmvn(m, mu = rep(0, p), sigma = sigma.mat)
y_test <- x_test %*% true.beta + rnorm(m, 0, sigma)

# Simulation of uncontaminated data 
x <- mvnfast::rmvn(n, mu = rep(0, p), sigma = sigma.mat)
y <- x %*% true.beta + rnorm(n, 0, sigma)

# Contamination of data 
contamination_indices <- 1:floor(n*contamination.prop)
k_lev <- 2
k_slo <- 100
x_train <- x
y_train <- y
beta_cont <- true.beta
beta_cont[true.beta!=0] <- beta_cont[true.beta!=0]*(1 + k_slo)
beta_cont[true.beta==0] <- k_slo*max(abs(true.beta))
for(cont_id in contamination_indices){
 
 a <- runif(p, min = -1, max = 1)
 a <- a - as.numeric((1/p)*t(a) %*% rep(1, p))
  x_train[cont_id,] <- mvnfast::rmvn(1, rep(0, p), 0.1^2*diag(p)) + k_lev * a / 
                        as.numeric(sqrt(t(a) %*% solve(sigma.mat) %*% a))
  y_train[cont_id] <- t(x_train[cont_id,]) %*% beta_cont
}

# RMSS
rmss_fit <- RMSS(x = x_train, y = y_train,
                 n_models = 3,
                 h_grid = c(35), t_grid = c(6, 8, 10), u_grid = c(1:3),
                 initial_estimator = "robStepSplitReg",
                 tolerance = 1e-1,
                 max_iter = 1e3,
                 neighborhood_search = FALSE,
                 neighborhood_search_tolerance = 1e-1)
rmss_coefs <- coef(rmss_fit, 
                   h_ind = 1, t_ind = 2, u_ind = 1,
                   group_index = 1:rmss_fit$n_models)
sens_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/p.active
spec_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/sum(rmss_coefs[-1]!=0)
rmss_preds <- predict(rmss_fit, newx = x_test,
                      h_ind = 1, t_ind = 2, u_ind = 1,
                      group_index = 1:rmss_fit$n_models,
                      dynamic = FALSE)
rmss_mspe <- mean((y_test - rmss_preds)^2)/sigma^2

Coefficients for cv.RMSS Object

Description

coef.cv.RMSS returns the coefficients for a cv.RMSS object.

Usage

## S3 method for class 'cv.RMSS'
coef(
  object,
  h_ind = NULL,
  t_ind = NULL,
  u_ind = NULL,
  individual_models = FALSE,
  group_index = NULL,
  ...
)

Arguments

object

An object of class cv.RMSS.

h_ind

Index for robustness parameter.

t_ind

Index for sparsity parameter.

u_ind

Index for diversity parameter.

individual_models

Argument to determine whether the coefficients of each model are returned. Default is FALSE.

group_index

Groups included in the ensemble. Default setting includes all the groups.

...

Additional arguments for compatibility.

Value

The coefficients for the cv.RMSS object.

Author(s)

Anthony-Alexander Christidis, anthony.christidis@stat.ubc.ca

Examples

# Simulation parameters
n <- 50
p <- 100
rho <- 0.8
rho.inactive <- 0.2
group.size <- 5
p.active <- 15
snr <- 2
contamination.prop <- 0.3

# Setting the seed
set.seed(0)

# Block Correlation
sigma.mat <- matrix(0, p, p)
sigma.mat[1:p.active, 1:p.active] <- rho.inactive
for(group in 0:(p.active/group.size - 1))
  sigma.mat[(group*group.size+1):(group*group.size+group.size),
            (group*group.size+1):(group*group.size+group.size)] <- rho
diag(sigma.mat) <- 1

# Simulation of beta vector
true.beta <- c(runif(p.active, 0, 5)*(-1)^rbinom(p.active, 1, 0.7), 
               rep(0, p - p.active))

# Setting the SD of the variance
sigma <- as.numeric(sqrt(t(true.beta) %*% sigma.mat %*% true.beta)/sqrt(snr))

# Simulation of test data
m <- 2e3
x_test <- mvnfast::rmvn(m, mu = rep(0, p), sigma = sigma.mat)
y_test <- x_test %*% true.beta + rnorm(m, 0, sigma)

# Simulation of uncontaminated data 
x <- mvnfast::rmvn(n, mu = rep(0, p), sigma = sigma.mat)
y <- x %*% true.beta + rnorm(n, 0, sigma)

# Contamination of data 
contamination_indices <- 1:floor(n*contamination.prop)
k_lev <- 2
k_slo <- 100
x_train <- x
y_train <- y
beta_cont <- true.beta
beta_cont[true.beta!=0] <- beta_cont[true.beta!=0]*(1 + k_slo)
beta_cont[true.beta==0] <- k_slo*max(abs(true.beta))
for(cont_id in contamination_indices){
 
 a <- runif(p, min = -1, max = 1)
 a <- a - as.numeric((1/p)*t(a) %*% rep(1, p))
  x_train[cont_id,] <- mvnfast::rmvn(1, rep(0, p), 0.1^2*diag(p)) + k_lev * a / 
                        as.numeric(sqrt(t(a) %*% solve(sigma.mat) %*% a))
  y_train[cont_id] <- t(x_train[cont_id,]) %*% beta_cont
}

# CV RMSS
rmss_fit <- cv.RMSS(x = x_train, y = y_train,
                    n_models = 3,
                    h_grid = c(35), t_grid = c(6, 8, 10), u_grid = c(1:3),
                    initial_estimator = "robStepSplitReg",
                    tolerance = 1e-1,
                    max_iter = 1e3,
                    neighborhood_search = FALSE,
                    neighborhood_search_tolerance = 1e-1,
                    n_folds = 5,
                    alpha = 1/4,
                    gamma = 1, 
                    n_threads = 1)
rmss_coefs <- coef(rmss_fit, 
                   h_ind = rmss_fit$h_opt, 
                   t_ind = rmss_fit$t_opt, 
                   u_ind = rmss_fit$u_opt,
                   group_index = 1:rmss_fit$n_models)
sens_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/p.active
spec_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/sum(rmss_coefs[-1]!=0)
rmss_preds <- predict(rmss_fit, newx = x_test,
                      h_ind = rmss_fit$h_opt, 
                      t_ind = rmss_fit$t_opt, 
                      u_ind = rmss_fit$u_opt,
                      group_index = 1:rmss_fit$n_models,
                      dynamic = FALSE)
rmss_mspe <- mean((y_test - rmss_preds)^2)/sigma^2

Cross-Validatoin for Robust Multi-Model Subset Selection

Description

cv.RMSS performs the cross-validation procedure for robust multi-model subset selection.

Usage

cv.RMSS(
  x,
  y,
  n_models,
  h_grid,
  t_grid,
  u_grid,
  initial_estimator = c("robStepSplitReg", "srlars")[1],
  tolerance = 0.1,
  max_iter = 1000,
  neighborhood_search = FALSE,
  neighborhood_search_tolerance = 0.1,
  cv_criterion = c("tau", "trimmed")[1],
  n_folds = 5,
  alpha = 1/4,
  gamma = 1,
  n_threads = 1
)

Arguments

x

Design matrix.

y

Response vector.

n_models

Number of models into which the variables are split.

h_grid

Grid for robustness parameter.

t_grid

Grid for sparsity parameter.

u_grid

Grid for diversity parameter.

initial_estimator

Method used for initial estimator. Must be one of "robStepSplitReg" (default) or "srlars".

tolerance

Tolerance level for convergence of PSBGD algorithm.

max_iter

Maximum number of iterations in PSBGD algorithm.

neighborhood_search

Neighborhood search to improve solution. Default is FALSE.

neighborhood_search_tolerance

Tolerance parameter for neighborhood search. Default is 1e-1.

cv_criterion

Criterion to use for cross-validation procedure. Must be one of "tau" (default) or "trimmed".

n_folds

Number of folds for cross-validation procedure. Default is 5.

alpha

Proportion of trimmed samples for cross-validation procedure. Default is 1/4.

gamma

Weight parameter for ensemble MSPE (gamma) and average MSPE of individual models (1 - gamma). Default is 1.

n_threads

Number of threads used by OpenMP for multithreading over the folds. Default is 1.

Value

An object of class cv.RMSS

Author(s)

Anthony-Alexander Christidis, anthony.christidis@stat.ubc.ca

Examples

# Simulation parameters
n <- 50
p <- 100
rho <- 0.8
rho.inactive <- 0.2
group.size <- 5
p.active <- 15
snr <- 2
contamination.prop <- 0.3

# Setting the seed
set.seed(0)

# Block Correlation
sigma.mat <- matrix(0, p, p)
sigma.mat[1:p.active, 1:p.active] <- rho.inactive
for(group in 0:(p.active/group.size - 1))
  sigma.mat[(group*group.size+1):(group*group.size+group.size),
            (group*group.size+1):(group*group.size+group.size)] <- rho
diag(sigma.mat) <- 1

# Simulation of beta vector
true.beta <- c(runif(p.active, 0, 5)*(-1)^rbinom(p.active, 1, 0.7), 
               rep(0, p - p.active))

# Setting the SD of the variance
sigma <- as.numeric(sqrt(t(true.beta) %*% sigma.mat %*% true.beta)/sqrt(snr))

# Simulation of test data
m <- 2e3
x_test <- mvnfast::rmvn(m, mu = rep(0, p), sigma = sigma.mat)
y_test <- x_test %*% true.beta + rnorm(m, 0, sigma)

# Simulation of uncontaminated data 
x <- mvnfast::rmvn(n, mu = rep(0, p), sigma = sigma.mat)
y <- x %*% true.beta + rnorm(n, 0, sigma)

# Contamination of data 
contamination_indices <- 1:floor(n*contamination.prop)
k_lev <- 2
k_slo <- 100
x_train <- x
y_train <- y
beta_cont <- true.beta
beta_cont[true.beta!=0] <- beta_cont[true.beta!=0]*(1 + k_slo)
beta_cont[true.beta==0] <- k_slo*max(abs(true.beta))
for(cont_id in contamination_indices){
 
 a <- runif(p, min = -1, max = 1)
 a <- a - as.numeric((1/p)*t(a) %*% rep(1, p))
  x_train[cont_id,] <- mvnfast::rmvn(1, rep(0, p), 0.1^2*diag(p)) + k_lev * a / 
                        as.numeric(sqrt(t(a) %*% solve(sigma.mat) %*% a))
  y_train[cont_id] <- t(x_train[cont_id,]) %*% beta_cont
}

# CV RMSS
rmss_fit <- cv.RMSS(x = x_train, y = y_train,
                    n_models = 3,
                    h_grid = c(35), t_grid = c(6, 8, 10), u_grid = c(1:3),
                    initial_estimator = "robStepSplitReg",
                    tolerance = 1e-1,
                    max_iter = 1e3,
                    neighborhood_search = FALSE,
                    neighborhood_search_tolerance = 1e-1,
                    n_folds = 5,
                    alpha = 1/4,
                    gamma = 1, 
                    n_threads = 1)
rmss_coefs <- coef(rmss_fit, 
                   h_ind = rmss_fit$h_opt, 
                   t_ind = rmss_fit$t_opt, 
                   u_ind = rmss_fit$u_opt,
                   group_index = 1:rmss_fit$n_models)
sens_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/p.active
spec_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/sum(rmss_coefs[-1]!=0)
rmss_preds <- predict(rmss_fit, newx = x_test,
                      h_ind = rmss_fit$h_opt, 
                      t_ind = rmss_fit$t_opt, 
                      u_ind = rmss_fit$u_opt,
                      group_index = 1:rmss_fit$n_models,
                      dynamic = FALSE)
rmss_mspe <- mean((y_test - rmss_preds)^2)/sigma^2

Predictions for RMSS Object

Description

predict.RMSS returns the predictions for a RMSS object.

Usage

## S3 method for class 'RMSS'
predict(
  object,
  newx,
  h_ind,
  t_ind,
  u_ind,
  group_index = NULL,
  dynamic = FALSE,
  ...
)

Arguments

object

An object of class RMSS.

newx

New data for predictions.

h_ind

Index for robustness parameter.

t_ind

Index for sparsity parameter.

u_ind

Index for diversity parameter.

group_index

Groups included in the ensemble. Default setting includes all the groups.

dynamic

Argument to determine whether dynamic predictions are used based on deviating cells. Default is FALSE.

...

Additional arguments for compatibility.

Value

The predictions for the RMSS object.

Author(s)

Anthony-Alexander Christidis, anthony.christidis@stat.ubc.ca

Examples

# Simulation parameters
n <- 50
p <- 100
rho <- 0.8
rho.inactive <- 0.2
group.size <- 5
p.active <- 15
snr <- 2
contamination.prop <- 0.3

# Setting the seed
set.seed(0)

# Block Correlation
sigma.mat <- matrix(0, p, p)
sigma.mat[1:p.active, 1:p.active] <- rho.inactive
for(group in 0:(p.active/group.size - 1))
  sigma.mat[(group*group.size+1):(group*group.size+group.size),
            (group*group.size+1):(group*group.size+group.size)] <- rho
diag(sigma.mat) <- 1

# Simulation of beta vector
true.beta <- c(runif(p.active, 0, 5)*(-1)^rbinom(p.active, 1, 0.7), 
               rep(0, p - p.active))

# Setting the SD of the variance
sigma <- as.numeric(sqrt(t(true.beta) %*% sigma.mat %*% true.beta)/sqrt(snr))

# Simulation of test data
m <- 2e3
x_test <- mvnfast::rmvn(m, mu = rep(0, p), sigma = sigma.mat)
y_test <- x_test %*% true.beta + rnorm(m, 0, sigma)

# Simulation of uncontaminated data 
x <- mvnfast::rmvn(n, mu = rep(0, p), sigma = sigma.mat)
y <- x %*% true.beta + rnorm(n, 0, sigma)

# Contamination of data 
contamination_indices <- 1:floor(n*contamination.prop)
k_lev <- 2
k_slo <- 100
x_train <- x
y_train <- y
beta_cont <- true.beta
beta_cont[true.beta!=0] <- beta_cont[true.beta!=0]*(1 + k_slo)
beta_cont[true.beta==0] <- k_slo*max(abs(true.beta))
for(cont_id in contamination_indices){
 
 a <- runif(p, min = -1, max = 1)
 a <- a - as.numeric((1/p)*t(a) %*% rep(1, p))
  x_train[cont_id,] <- mvnfast::rmvn(1, rep(0, p), 0.1^2*diag(p)) + k_lev * a / 
                        as.numeric(sqrt(t(a) %*% solve(sigma.mat) %*% a))
  y_train[cont_id] <- t(x_train[cont_id,]) %*% beta_cont
}

# RMSS
rmss_fit <- RMSS(x = x_train, y = y_train,
                 n_models = 3,
                 h_grid = c(35), t_grid = c(6, 8, 10), u_grid = c(1:3),
                 initial_estimator = "robStepSplitReg",
                 tolerance = 1e-1,
                 max_iter = 1e3,
                 neighborhood_search = FALSE,
                 neighborhood_search_tolerance = 1e-1)
rmss_coefs <- coef(rmss_fit, 
                   h_ind = 1, t_ind = 2, u_ind = 1,
                   group_index = 1:rmss_fit$n_models)
sens_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/p.active
spec_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/sum(rmss_coefs[-1]!=0)
rmss_preds <- predict(rmss_fit, newx = x_test,
                      h_ind = 1, t_ind = 2, u_ind = 1,
                      group_index = 1:rmss_fit$n_models,
                      dynamic = FALSE)
rmss_mspe <- mean((y_test - rmss_preds)^2)/sigma^2

Predictions for cv.RMSS Object

Description

predict.cv.RMSS returns the predictions for a cv.RMSS object.

Usage

## S3 method for class 'cv.RMSS'
predict(
  object,
  newx,
  h_ind = NULL,
  t_ind = NULL,
  u_ind = NULL,
  group_index = NULL,
  dynamic = FALSE,
  ...
)

Arguments

object

An object of class cv.RMSS.

newx

New data for predictions.

h_ind

Index for robustness parameter.

t_ind

Index for sparsity parameter.

u_ind

Index for diversity parameter.

group_index

Groups included in the ensemble. Default setting includes all the groups.

dynamic

Argument to determine whether dynamic predictions are used based on deviating cells. Default is FALSE.

...

Additional arguments for compatibility.

Value

The predictions for the cv.RMSS object.

Author(s)

Anthony-Alexander Christidis, anthony.christidis@stat.ubc.ca

Examples

# Simulation parameters
n <- 50
p <- 100
rho <- 0.8
rho.inactive <- 0.2
group.size <- 5
p.active <- 15
snr <- 2
contamination.prop <- 0.3

# Setting the seed
set.seed(0)

# Block Correlation
sigma.mat <- matrix(0, p, p)
sigma.mat[1:p.active, 1:p.active] <- rho.inactive
for(group in 0:(p.active/group.size - 1))
  sigma.mat[(group*group.size+1):(group*group.size+group.size),
            (group*group.size+1):(group*group.size+group.size)] <- rho
diag(sigma.mat) <- 1

# Simulation of beta vector
true.beta <- c(runif(p.active, 0, 5)*(-1)^rbinom(p.active, 1, 0.7), 
               rep(0, p - p.active))

# Setting the SD of the variance
sigma <- as.numeric(sqrt(t(true.beta) %*% sigma.mat %*% true.beta)/sqrt(snr))

# Simulation of test data
m <- 2e3
x_test <- mvnfast::rmvn(m, mu = rep(0, p), sigma = sigma.mat)
y_test <- x_test %*% true.beta + rnorm(m, 0, sigma)

# Simulation of uncontaminated data 
x <- mvnfast::rmvn(n, mu = rep(0, p), sigma = sigma.mat)
y <- x %*% true.beta + rnorm(n, 0, sigma)

# Contamination of data 
contamination_indices <- 1:floor(n*contamination.prop)
k_lev <- 2
k_slo <- 100
x_train <- x
y_train <- y
beta_cont <- true.beta
beta_cont[true.beta!=0] <- beta_cont[true.beta!=0]*(1 + k_slo)
beta_cont[true.beta==0] <- k_slo*max(abs(true.beta))
for(cont_id in contamination_indices){
 
 a <- runif(p, min = -1, max = 1)
 a <- a - as.numeric((1/p)*t(a) %*% rep(1, p))
  x_train[cont_id,] <- mvnfast::rmvn(1, rep(0, p), 0.1^2*diag(p)) + k_lev * a / 
                        as.numeric(sqrt(t(a) %*% solve(sigma.mat) %*% a))
  y_train[cont_id] <- t(x_train[cont_id,]) %*% beta_cont
}

# CV RMSS
rmss_fit <- cv.RMSS(x = x_train, y = y_train,
                    n_models = 3,
                    h_grid = c(35), t_grid = c(6, 8, 10), u_grid = c(1:3),
                    initial_estimator = "robStepSplitReg",
                    tolerance = 1e-1,
                    max_iter = 1e3,
                    neighborhood_search = FALSE,
                    neighborhood_search_tolerance = 1e-1,
                    n_folds = 5,
                    alpha = 1/4,
                    gamma = 1, 
                    n_threads = 1)
rmss_coefs <- coef(rmss_fit, 
                   h_ind = rmss_fit$h_opt, 
                   t_ind = rmss_fit$t_opt, 
                   u_ind = rmss_fit$u_opt,
                   group_index = 1:rmss_fit$n_models)
sens_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/p.active
spec_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/sum(rmss_coefs[-1]!=0)
rmss_preds <- predict(rmss_fit, newx = x_test,
                      h_ind = rmss_fit$h_opt, 
                      t_ind = rmss_fit$t_opt, 
                      u_ind = rmss_fit$u_opt,
                      group_index = 1:rmss_fit$n_models,
                      dynamic = FALSE)
rmss_mspe <- mean((y_test - rmss_preds)^2)/sigma^2

Trimmed samples for RMSS or cv.RMSS Object

Description

trimmed_samples returns the coefficients for a RMSS or cv.RMSS object.

Usage

trimmed_samples(
  object,
  h_ind = NULL,
  t_ind = NULL,
  u_ind = NULL,
  group_index = NULL,
  ...
)

Arguments

object

An object of class RMSS

h_ind

Index for robustness parameter.

t_ind

Index for sparsity parameter.

u_ind

Index for diversity parameter.

group_index

Groups included in the ensemble. Default setting includes all the groups.

...

Additional arguments for compatibility.

Value

The trimmed samples for the RMSS or cv.RMSS object.

Author(s)

Anthony-Alexander Christidis, anthony.christidis@stat.ubc.ca

Examples

# Simulation parameters
n <- 50
p <- 100
rho <- 0.8
rho.inactive <- 0.2
group.size <- 5
p.active <- 15
snr <- 2
contamination.prop <- 0.3

# Setting the seed
set.seed(0)

# Block Correlation
sigma.mat <- matrix(0, p, p)
sigma.mat[1:p.active, 1:p.active] <- rho.inactive
for(group in 0:(p.active/group.size - 1))
  sigma.mat[(group*group.size+1):(group*group.size+group.size),
            (group*group.size+1):(group*group.size+group.size)] <- rho
diag(sigma.mat) <- 1

# Simulation of beta vector
true.beta <- c(runif(p.active, 0, 5)*(-1)^rbinom(p.active, 1, 0.7), 
               rep(0, p - p.active))

# Setting the SD of the variance
sigma <- as.numeric(sqrt(t(true.beta) %*% sigma.mat %*% true.beta)/sqrt(snr))

# Simulation of test data
m <- 2e3
x_test <- mvnfast::rmvn(m, mu = rep(0, p), sigma = sigma.mat)
y_test <- x_test %*% true.beta + rnorm(m, 0, sigma)

# Simulation of uncontaminated data 
x <- mvnfast::rmvn(n, mu = rep(0, p), sigma = sigma.mat)
y <- x %*% true.beta + rnorm(n, 0, sigma)

# Contamination of data 
contamination_indices <- 1:floor(n*contamination.prop)
k_lev <- 2
k_slo <- 100
x_train <- x
y_train <- y
beta_cont <- true.beta
beta_cont[true.beta!=0] <- beta_cont[true.beta!=0]*(1 + k_slo)
beta_cont[true.beta==0] <- k_slo*max(abs(true.beta))
for(cont_id in contamination_indices){
 
 a <- runif(p, min = -1, max = 1)
 a <- a - as.numeric((1/p)*t(a) %*% rep(1, p))
  x_train[cont_id,] <- mvnfast::rmvn(1, rep(0, p), 0.1^2*diag(p)) + k_lev * a / 
                        as.numeric(sqrt(t(a) %*% solve(sigma.mat) %*% a))
  y_train[cont_id] <- t(x_train[cont_id,]) %*% beta_cont
}

# RMSS
rmss_fit <- RMSS(x = x_train, y = y_train,
                 n_models = 3,
                 h_grid = c(35), t_grid = c(6, 8, 10), u_grid = c(1:3),
                 tolerance = 1e-1,
                 max_iter = 1e3,
                 neighborhood_search = FALSE,
                 neighborhood_search_tolerance = 1e-1)
rmss_coefs <- coef(rmss_fit, 
                   h_ind = 1, t_ind = 2, u_ind = 1,
                   group_index = 1:rmss_fit$n_models)
sens_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/p.active
spec_rmss <- sum(which((rmss_coefs[-1]!=0)) <= p.active)/sum(rmss_coefs[-1]!=0)
rmss_preds <- predict(rmss_fit, newx = x_test,
                      h_ind = 1, t_ind = 2, u_ind = 1,
                      group_index = 1:rmss_fit$n_models,
                      dynamic = FALSE)
rmss_mspe <- mean((y_test - rmss_preds)^2)/sigma^2
trimmed_id <- trimmed_samples(rmss_fit, h_ind = 1, t_ind = 1, u_ind = 1)

Robust Multi-Model Subset Selection

Description

Usage

Arguments

Value

Author(s)

See Also

Examples

Coefficients for RMSS Object

Description

Usage

Arguments

Value

Author(s)

See Also

Examples

Coefficients for cv.RMSS Object

Description

Usage

Arguments

Value

Author(s)

See Also

Examples

Cross-Validatoin for Robust Multi-Model Subset Selection

Description

Usage

Arguments

Value

Author(s)

See Also

Examples

Predictions for RMSS Object

Description

Usage

Arguments

Value

Author(s)

See Also

Examples

Predictions for cv.RMSS Object

Description

Usage

Arguments

Value

Author(s)

See Also

Examples

Trimmed samples for RMSS or cv.RMSS Object

Description

Usage

Arguments

Value

Author(s)

See Also

Examples