def experiment_schizophrenia_data(data_path='data', n_folds=5,
iterations=10000,
verbose=True, plot=True, random_state=None):
"""Run the experiments on the Schizophrenia dataset
Parameters:
----------
data_path: string
Path to the folder containing the dataset.
n_folds: int
The number of folds in a StratifiedKFold cross-validation
iterations: int
Number of iterations to compute the null distribution of
balanced_accuracy and MMD^2_u
verbose: bool
plot: bool
Whether to plot the results of the statistical tests.
"""
name = 'Schizophrenia'
if verbose:
print '\nWorking on %s dataset...' % name
print '-----------------------'
X, y = load_schizophrenia_data(data_path, verbose=verbose)
# DCE + RBF
if verbose:
print '\n### Results for DCE_Embedding ###'
gk_dce = GK_DCE(kernel_vector_space='rbf')
K_dce = gk_dce.compare_pairwise(X)
simple_experiment(K_dce, y, n_folds=n_folds, iterations=iterations,
verbose=verbose, data_name=name + '_dce', plot=plot,
random_state=random_state)
# DRE + RBF
if verbose:
print '\n### Results for DR_Embedding ###'
gk_dre = GK_DRE(kernel_vector_space='rbf')
K_dre = gk_dre.compare_pairwise(X)
simple_experiment(K_dre, y, n_folds=n_folds, iterations=iterations,
verbose=verbose, data_name=name+'_dr',plot=plot,
random_state=random_state)
# WL Kernel
if verbose:
print '\n### Results for WL_K_Embedding ###'
gk_wl = GK_WL(th=0.2)
K_wl = gk_wl.compare_pairwise(X)
simple_experiment(K_wl, y, n_folds=n_folds,
iterations=iterations,
verbose=verbose, data_name=name+'_wl', plot=plot,
random_state=random_state)
# SP Kernel
if verbose:
print '\n### Results for SP_K_Embedding ###'
gk_sp = GK_SP(th=0.2)
K_sp = gk_sp.compare_pairwise(X)
simple_experiment(K_sp, y, n_folds=n_folds,
iterations=iterations,
verbose=verbose, data_name=name+'_sp', plot=plot,
random_state=random_state)
#
# NBS
th = 0.5
mxcmp, mxcmp_null, p_value = apply_nbs(X, y, th, iterations, verbose)
if plot:
plot_statistic(mxcmp, mxcmp_null, p_value, data_name=name+'_nbs',
stats_name='Max_comp_size')
def experiment_schizophrenia_data(data_path='data',
n_folds=5,
iterations=10000,
verbose=True,
plot=True,
random_state=None):
"""Run the experiments on the Schizophrenia dataset
Parameters:
----------
data_path: string
Path to the folder containing the dataset.
n_folds: int
The number of folds in a StratifiedKFold cross-validation
iterations: int
Number of iterations to compute the null distribution of
balanced_accuracy and MMD^2_u
verbose: bool
plot: bool
Whether to plot the results of the statistical tests.
"""
name = 'Schizophrenia'
if verbose:
print '\nWorking on %s dataset...' % name
print '-----------------------'
X, y = load_schizophrenia_data(data_path, verbose=verbose)
# DCE Embedding
if verbose:
print '\n### Results for DCE_Embedding ###'
X_dce = DCE_embedding(X)
K_dce = compute_rbf_kernel_matrix(X_dce)
simple_experiment(K_dce,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_dce',
plot=plot,
random_state=random_state)
# DR Embedding
if verbose:
print '\n### Results for DR_Embedding ###'
X_dr = DR_embedding(X)
K_dr = compute_rbf_kernel_matrix(X_dr)
simple_experiment(K_dr,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_dr',
plot=plot,
random_state=random_state)
# WL Kernel based Embedding
if verbose:
print '\n### Results for WL_K_Embedding ###'
th = 0.2
K_wl = WL_K_embedding(X, th)
simple_experiment(K_wl,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_wl',
plot=plot,
random_state=random_state)
# SP Kernel based Embedding
if verbose:
print '\n### Results for SP_K_Embedding ###'
th = 0.2
K_sp = SP_K_embedding(X, th)
simple_experiment(K_sp,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_sp',
plot=plot,
random_state=random_state)
# for name in locs:
# if verbose:
# print '\nWorking on %s dataset...' % name
# print '-----------------------'
# X, y = load_1000_funct_connectome(data_path, name,
# verbose=verbose)
# check_instability_classification(X, y, location=name,
# n_folds=n_folds,
# iterations=iterations,
# verbose=verbose,
# reps=repetitions, seed=seed)
# Working on the schizophrenia data
if verbose:
print '\nWorking on Schizophrenia dataset...'
X, y = load_schizophrenia_data(data_path, verbose=verbose)
check_instability_classification(X,
y,
location='Schizophrenia',
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
reps=repetitions,
seed=seed)
# # Working on URI data
# X, y = load_kernel_matrix()
# check_instability_classification(X, y, n_folds=n_folds,
# iterations=iterations,
# verbose=verbose,
# reps=repetitions, seed=seed)
def experiment_schizophrenia_data(data_path='data', n_folds=5,
iterations=10000,
verbose=True, plot=True, random_state=None):
"""Run the experiments on the Schizophrenia dataset
Parameters:
----------
data_path: string
Path to the folder containing the dataset.
n_folds: int
The number of folds in a StratifiedKFold cross-validation
iterations: int
Number of iterations to compute the null distribution of
balanced_accuracy and MMD^2_u
verbose: bool
plot: bool
Whether to plot the results of the statistical tests.
"""
name = 'Schizophrenia'
if verbose:
print '\nWorking on %s dataset...' % name
print '-----------------------'
X, y = load_schizophrenia_data(data_path, verbose=verbose)
# DCE Embedding
if verbose:
print '\n### Results for DCE_Embedding ###'
X_dce = DCE_embedding(X)
K_dce = compute_rbf_kernel_matrix(X_dce)
simple_experiment(K_dce, y, n_folds=n_folds, iterations=iterations,
verbose=verbose, data_name=name + '_dce', plot=plot,
random_state=random_state)
# DR Embedding
if verbose:
print '\n### Results for DR_Embedding ###'
X_dr = DR_embedding(X)
K_dr = compute_rbf_kernel_matrix(X_dr)
simple_experiment(K_dr, y, n_folds=n_folds, iterations=iterations,
verbose=verbose, data_name=name+'_dr',plot=plot,
random_state=random_state)
# WL Kernel based Embedding
if verbose:
print '\n### Results for WL_K_Embedding ###'
th = 0.2
K_wl = WL_K_embedding(X, th)
simple_experiment(K_wl, y, n_folds=n_folds,
iterations=iterations,
verbose=verbose, data_name=name+'_wl', plot=plot,
random_state=random_state)
# SP Kernel based Embedding
if verbose:
print '\n### Results for SP_K_Embedding ###'
th = 0.2
K_sp = SP_K_embedding(X, th)
simple_experiment(K_sp, y, n_folds=n_folds,
iterations=iterations,
verbose=verbose, data_name=name+'_sp', plot=plot,
random_state=random_state)
# # Working on the functional connectome data
# for name in locs:
# if verbose:
# print '\nWorking on %s dataset...' % name
# print '-----------------------'
# X, y = load_1000_funct_connectome(data_path, name,
# verbose=verbose)
# check_instability_classification(X, y, location=name,
# n_folds=n_folds,
# iterations=iterations,
# verbose=verbose,
# reps=repetitions, seed=seed)
# Working on the schizophrenia data
if verbose:
print '\nWorking on Schizophrenia dataset...'
X, y = load_schizophrenia_data(data_path, verbose=verbose)
check_instability_classification(X, y, location='Schizophrenia',
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
reps=repetitions, seed=seed)
# # Working on URI data
# X, y = load_kernel_matrix()
# check_instability_classification(X, y, n_folds=n_folds,
# iterations=iterations,
# verbose=verbose,
# reps=repetitions, seed=seed)
def experiment_schizophrenia_data(data_path='data',
n_folds=5,
iterations=10000,
verbose=True,
plot=True,
random_state=None):
"""Run the experiments on the Schizophrenia dataset
Parameters:
----------
data_path: string
Path to the folder containing the dataset.
n_folds: int
The number of folds in a StratifiedKFold cross-validation
iterations: int
Number of iterations to compute the null distribution of
balanced_accuracy and MMD^2_u
verbose: bool
plot: bool
Whether to plot the results of the statistical tests.
"""
name = 'Schizophrenia'
if verbose:
print '\nWorking on %s dataset...' % name
print '-----------------------'
X, y = load_schizophrenia_data(data_path, verbose=verbose)
# DCE + RBF
if verbose:
print '\n### Results for DCE_Embedding ###'
gk_dce = GK_DCE(kernel_vector_space='rbf')
K_dce = gk_dce.compare_pairwise(X)
simple_experiment(K_dce,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_dce',
plot=plot,
random_state=random_state)
# DRE + RBF
if verbose:
print '\n### Results for DR_Embedding ###'
gk_dre = GK_DRE(kernel_vector_space='rbf')
K_dre = gk_dre.compare_pairwise(X)
simple_experiment(K_dre,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_dr',
plot=plot,
random_state=random_state)
# WL Kernel
if verbose:
print '\n### Results for WL_K_Embedding ###'
gk_wl = GK_WL(th=0.2)
K_wl = gk_wl.compare_pairwise(X)
simple_experiment(K_wl,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_wl',
plot=plot,
random_state=random_state)
# SP Kernel
if verbose:
print '\n### Results for SP_K_Embedding ###'
gk_sp = GK_SP(th=0.2)
K_sp = gk_sp.compare_pairwise(X)
simple_experiment(K_sp,
y,
n_folds=n_folds,
iterations=iterations,
verbose=verbose,
data_name=name + '_sp',
plot=plot,
random_state=random_state)
#
# NBS
th = 0.5
mxcmp, mxcmp_null, p_value = apply_nbs(X, y, th, iterations, verbose)
if plot:
plot_statistic(mxcmp,
mxcmp_null,
p_value,
data_name=name + '_nbs',
stats_name='Max_comp_size')
