FIX: wrong size specification in bootstrap (mne-tools#4666)

* wrong size specification in bootstrap

* stat_fun

examples/time_frequency/plot_time_frequency_global_field_power.py

@@ -113,7 +113,7 @@
     ax.plot(times, gfp, label=freq_name, color=color, linewidth=2.5)
     ax.axhline(0, linestyle='--', color='grey', linewidth=2)
     ci_low, ci_up = _bootstrap_ci(average.data, random_state=0,
-                                  statfun=lambda x: np.sum(x ** 2, axis=0))
+                                  stat_fun=lambda x: np.sum(x ** 2, axis=0))
     ci_low = rescale(ci_low, average.times, baseline=(None, 0))
     ci_up = rescale(ci_up, average.times, baseline=(None, 0))
     ax.fill_between(times, gfp + ci_up, gfp - ci_low, color=color, alpha=0.3)

mne/stats/permutations.py

@@ -151,21 +151,21 @@ def permutation_t_test(X, n_permutations=10000, tail=0, n_jobs=1,
     return T_obs, p_values, H0
 
 
-def _bootstrap_ci(arr, ci=.95, n_bootstraps=2000, statfun='mean',
+def _bootstrap_ci(arr, ci=.95, n_bootstraps=2000, stat_fun='mean',
                   random_state=None):
     """Get confidence intervals from non-parametric bootstrap."""
-    if statfun == "mean":
-        statfun = lambda x: x.mean(axis=0)  # noqa
-    elif statfun == 'median':
-        statfun = lambda x: np.median(x, axis=0)  # noqa
-    elif not callable(statfun):
-        raise ValueError("statfun must be 'mean', 'median' or callable.")
+    if stat_fun == "mean":
+        stat_fun = lambda x: x.mean(axis=0)  # noqa
+    elif stat_fun == 'median':
+        stat_fun = lambda x: np.median(x, axis=0)  # noqa
+    elif not callable(stat_fun):
+        raise ValueError("stat_fun must be 'mean', 'median' or callable.")
     n_trials = arr.shape[0]
     indices = np.arange(n_trials, dtype=int)  # BCA would be cool to have too
     rng = check_random_state(random_state)
     boot_indices = rng.choice(indices, replace=True,
-                              size=(n_trials, len(indices)))
-    stat = np.array([statfun(arr[inds]) for inds in boot_indices])
+                              size=(n_bootstraps, len(indices)))
+    stat = np.array([stat_fun(arr[inds]) for inds in boot_indices])
     ci = (((1 - ci) / 2) * 100, ((1 - ((1 - ci) / 2))) * 100)
     ci_low, ci_up = np.percentile(stat, ci, axis=0)
     return np.array([ci_low, ci_up])

mne/stats/tests/test_permutations.py

@@ -43,8 +43,8 @@ def test_ci():
     arr = np.linspace(0, 1, 1000)[..., np.newaxis]
     assert_allclose(_ci(arr, method="parametric"),
                     _ci(arr, method="bootstrap"), rtol=.005)
-    assert_allclose(_bootstrap_ci(arr, statfun="median", random_state=0),
-                    _bootstrap_ci(arr, statfun="mean", random_state=0),
+    assert_allclose(_bootstrap_ci(arr, stat_fun="median", random_state=0),
+                    _bootstrap_ci(arr, stat_fun="mean", random_state=0),
                     rtol=.1)
 
 run_tests_if_main()