Add tools for testing and make an example with CPI

examples/plot_2D_simulation_example.py

@@ -69,7 +69,7 @@
     ensemble_clustered_inference_pvalue,
 )
 from hidimstat.statistical_tools.p_values import zscore_from_pval
-from hidimstat._utils.scenario import multivariate_simulation
+from hidimstat._utils.scenario import multivariate_simulation_spatial
 
 #############################################################################
 # Specific plotting functions
@@ -171,7 +171,7 @@ def plot(maps, titles):
 smooth_X = 1.0  # level of spatial smoothing introduced by the Gaussian filter
 
 # generating the data
-X_init, y, beta, epsilon, _, _ = multivariate_simulation(
+X_init, y, beta, epsilon, _, _ = multivariate_simulation_spatial(
     n_samples, shape, roi_size, sigma, smooth_X, seed=1
 )
 

examples/plot_dcrt_example.py

@@ -15,7 +15,7 @@
 import matplotlib.pyplot as plt
 import numpy as np
 from hidimstat.dcrt import dcrt_zero, dcrt_pvalue
-from hidimstat._utils.scenario import multivariate_1D_simulation
+from hidimstat._utils.scenario import multivariate_simulation
 
 plt.rcParams.update({"font.size": 21})
 
@@ -43,8 +43,14 @@
     # Nominal false positive rate
     alpha = 5e-2
 
-    X, y, _, __ = multivariate_1D_simulation(
-        n_samples=n, n_features=p, support_size=n_signal, rho=rho, seed=sim_ind
+    X, y, beta_true, _, _, _ = multivariate_simulation(
+        n_samples=n,
+        n_features=p,
+        support_size=n_signal,
+        rho=rho,
+        snr=snr,
+        shuffle=True,
+        seed=sim_ind,
     )
 
     # Applying a reLu function on the outcome y to get non-linear relationships
@@ -55,8 +61,10 @@
     variables_important_lasso, pvals_lasso, ts_lasso = dcrt_pvalue(
         selection_features, X_res, sigma2, y_res
     )
-    typeI_error["Lasso"].append(sum(pvals_lasso[n_signal:] < alpha) / (p - n_signal))
-    power["Lasso"].append(sum(pvals_lasso[:n_signal] < alpha) / (n_signal))
+    typeI_error["Lasso"].append(
+        sum(pvals_lasso[np.logical_not(beta_true)] < alpha) / (p - n_signal)
+    )
+    power["Lasso"].append(sum(pvals_lasso[beta_true] < alpha) / (n_signal))
 
     ## dcrt Random Forest ##
     selection_features, X_res, sigma2, y_res = dcrt_zero(
@@ -65,8 +73,10 @@
     rvariables_important_forest, pvals_forest, ts_forest = dcrt_pvalue(
         selection_features, X_res, sigma2, y_res
     )
-    typeI_error["Forest"].append(sum(pvals_forest[n_signal:] < alpha) / (p - n_signal))
-    power["Forest"].append(sum(pvals_forest[:n_signal] < alpha) / (n_signal))
+    typeI_error["Forest"].append(
+        sum(pvals_forest[np.logical_not(beta_true)] < alpha) / (p - n_signal)
+    )
+    power["Forest"].append(sum(pvals_forest[beta_true] < alpha) / (n_signal))
 
 #############################################################################
 # Plotting the comparison

examples/plot_importance_classification_iris.py

@@ -116,7 +116,7 @@ def run_one_fold(X, y, model, train_index, test_index, vim_name="CPI", groups=No
     GridSearchCV(SVC(kernel="rbf"), {"C": np.logspace(-3, 3, 10)}),
 ]
 cv = KFold(n_splits=5, shuffle=True, random_state=0)
-groups = {ft: i for i, ft in enumerate(dataset.feature_names)}
+groups = {ft: [i] for i, ft in enumerate(dataset.feature_names)}
 out_list = Parallel(n_jobs=5)(
     delayed(run_one_fold)(
         X, y, model, train_index, test_index, vim_name=vim_name, groups=groups

examples/plot_knockoff_aggregation.py

@@ -31,7 +31,7 @@
     model_x_knockoff_pvalue,
 )
 from hidimstat.statistical_tools.multiple_testing import fdp_power
-from hidimstat._utils.scenario import multivariate_1D_simulation_AR
+from hidimstat._utils.scenario import multivariate_simulation
 
 
 #############################################################################
@@ -75,9 +75,15 @@
 # ---------------------------------------------------------------------
 def single_run(n_samples, n_features, rho, sparsity, snr, fdr, n_bootstraps, seed=None):
     # Generate data
-    X, y, _, non_zero_index = multivariate_1D_simulation_AR(
-        n_samples, n_features, rho=rho, sparsity=sparsity, seed=seed, snr=snr
+    X, y, beta_true, _, _, _ = multivariate_simulation(
+        n_samples,
+        n_features,
+        rho=rho,
+        support_size=int(n_features * sparsity),
+        snr=snr,
+        seed=seed,
     )
+    non_zero_index = np.where(beta_true)[0]
 
     # Use model-X Knockoffs [1]
     selected, test_scores, threshold, X_tildes = model_x_knockoff(

pyproject.toml

@@ -49,13 +49,8 @@ doc = [
     "sphinx-prompt          >= 1.0.0",
     "seaborn",
 ]
-plotting = [
-    "matplotlib>=3.9.0",
-]
-style = [
-    "black  >= 24.4.2", 
-    "isort  >= 5.13.2"
-]
+plotting = ["matplotlib>=3.9.0"]
+style = ["black  >= 24.4.2", "isort  >= 5.13.2"]
 # For running unit and docstring tests
 test = [
     "iniconfig  >= 0.1,  < 3",
@@ -67,7 +62,8 @@ test = [
     "pytest-xdist[psutil]",
     "pytest-html",
     "pytest-timeout",
-    "pytest-durations"
+    "pytest-durations",
+    "pytest-env",
 ]
 
 [project.urls]
@@ -96,20 +92,26 @@ minversion = "8.0"
 pythonpath = "src"
 testpaths = "test"
 addopts = [
-    "-ra",                   # short test summary info
+    "-ra",                     # short test summary info
     "--import-mode=importlib", # better control of importing packages
-    "--showlocals",          # show local variable inn trackbacks
-    "--strict-config",       # no warning from parsing pytest configuration file
-    "--strict-markers",      # undefine markers will raise an error
+    "--showlocals",            # show local variable inn trackbacks
+    "--strict-config",         # no warning from parsing pytest configuration file
+    "--strict-markers",        # undefine markers will raise an error
     # pytest-randomly
     # "--randomly-dont-reset-seed", # turn off the rest of the random seed 
     # pytest-xdist option
-    "-n=auto",               # automatically define the number of process
+    "-n=auto",                  # automatically define the number of process
 ]
 markers = [
     "slow: marks tests as slow (deselect with '-m \"not slow\"')",
 ]
 # pytest-timeout
-timeout= 60          # on individual test should not take more than 10s
-session_timeout=1200 # all the tests should be run in 5 min
+timeout = 60           # on individual test should not take more than 10s
+session_timeout = 1200 # all the tests should be run in 5 min
+
 
+[tool.pytest_env]
+OPENBLAS_NUM_THREADS = 1
+BLIS_NUM_THREADS = 1
+MKL_NUM_THREADS = 1
+OMP_NUM_THREADS = 1