Fixed a bunch of codacy issues.

README.md

@@ -2,8 +2,6 @@
 
 [![Codacy Badge](https://api.codacy.com/project/badge/Grade/963ddefc14884a289b800497e74c4e45)](https://app.codacy.com/gh/courtois-neuromod/dypac?utm_source=github.com&utm_medium=referral&utm_content=courtois-neuromod/dypac&utm_campaign=Badge_Grade_Dashboard) [![CircleCI](https://circleci.com/gh/courtois-neuromod/dypac.svg?style=svg)](https://circleci.com/gh/courtois-neuromod/dypac) [![codecov](https://codecov.io/gh/courtois-neuromod/dypac/branch/master/graph/badge.svg)](https://codecov.io/gh/courtois-neuromod/dypac)
 
-
-
 Detecting stable dynamic parcellation in fMRI data on the full brain.
 
-The algorithm is a simple two level clustering, one on sliding time windows, and one on indicator functions of parcels agreggated over many windows. Optionally the approach can be iterated over several runs. 
+The algorithm is a simple two level clustering, one on sliding time windows, and one on indicator functions of parcels agreggated over many windows. Optionally the approach can be iterated over several runs.

dypac/__init__.py

@@ -1,6 +1,4 @@
-"""
-Dynamic Parcel Aggregation with Clustering (dypac)
-"""
+"""Dynamic Parcel Aggregation with Clustering (dypac)."""
 from .dypac import dypac
 from .bascpp import replicate_clusters, find_states, stab_maps
 __all__ = ['dypac', 'replicate_clusters', 'find_states', 'stab_maps']

dypac/bascpp.py

@@ -7,7 +7,7 @@
 # License: BSD 3 clause
 from tqdm import tqdm
 
-from scipy.sparse import csr_matrix, vstack, find
+from scipy.sparse import csr_matrix, find
 import numpy as np
 
 from sklearn.cluster import k_means
@@ -61,7 +61,7 @@ def _start_window(n_time, n_replications, subsample_size):
 def _trim_states(onehot, states, n_states, verbose, threshold_sim):
     """Trim the states clusters to exclude outliers."""
     for ss in tqdm(range(n_states), disable=not verbose, desc="Trimming states"):
-        [ix, iy, val] = find(onehot[states == ss, :])
+        ix, iy, _ = find(onehot[states == ss, :])
         size_onehot = np.array(onehot[states == ss, :].sum(axis=1)).flatten()
         ref_cluster = np.array(onehot[states == ss, :].mean(dtype="float", axis=0))
         avg_stab = np.bincount(ix, weights=ref_cluster[0,iy].flatten())
@@ -120,7 +120,7 @@ def replicate_clusters(
             samp = scale(samp, axis=1)
         if embedding.shape[0] > 0:
             samp = np.concatenate([samp, embedding], axis=1)
-        cent, part[rr, :], inert = k_means(
+        _, part[rr, :], _ = k_means(
             samp,
             n_clusters=n_clusters,
             init="k-means++",

dypac/dypac.py

@@ -23,7 +23,8 @@
 
 
 class dypac(BaseDecomposition):
-    """Perform Stable Dynamic Cluster Analysis.
+    """
+    Perform Stable Dynamic Cluster Analysis.
 
     Parameters
     ----------
@@ -130,7 +131,6 @@ class dypac(BaseDecomposition):
         See http://nilearn.github.io/manipulating_images/input_output.html
         The mask of the data. If no mask was given at masker creation, contains
         the automatically computed mask.
-
     """
 
     def __init__(
@@ -160,6 +160,7 @@ def __init__(
         memory_level=0,
         verbose=1,
     ):
+        """Set up default attributes for the class."""
         # All those settings are taken from nilearn BaseDecomposition
         self.random_state = random_state
         self.mask = mask
@@ -190,6 +191,7 @@ def __init__(
         self.threshold_sim = threshold_sim
 
     def _check_components_(self):
+        """Check for presence of estimated components."""
         if not hasattr(self, "components_"):
             raise ValueError(
                 "Object has no components_ attribute. "
@@ -198,7 +200,8 @@ def _check_components_(self):
             )
 
     def fit(self, imgs, confounds=None):
-        """Compute the mask and the dynamic parcels across datasets
+        """
+        Compute the mask and the dynamic parcels across datasets.
 
         Parameters
         ----------
@@ -217,7 +220,6 @@ def fit(self, imgs, confounds=None):
          self: object
             Returns the instance itself. Contains attributes listed
             at the object level.
-
         """
         # Base fit for decomposition estimators : compute the embedded masker
         if isinstance(imgs, str):
@@ -306,9 +308,12 @@ def _mask_and_reduce(self, imgs, confounds=None):
         stable dynamic parcels from a list of 4D fMRI datasets.
 
         Returns
-        ------
+        -------
         stab_maps: ndarray
-            Concatenation of dynamic parcels across all datasets.
+            stability maps of each state.
+
+        dwell_time: ndarray
+            dwell time of each state.
         """
 
         for ind, img, confound in zip(range(len(imgs)), imgs, confounds):
@@ -353,7 +358,7 @@ def _mask_and_reduce(self, imgs, confounds=None):
         return stab_maps, dwell_time
 
     def transform_sparse(self, img, confound=None):
-        """Transform a 4D dataset in a component space"""
+        """Transform a 4D dataset in a component space."""
         self._check_components_()
         this_data = self.masker_.transform(img, confound)
         del img
@@ -363,6 +368,6 @@ def transform_sparse(self, img, confound=None):
         return reg.coef_
 
     def inverse_transform_sparse(self, weights):
-        """Transform component weights as a 4D dataset"""
+        """Transform component weights as a 4D dataset."""
         self._check_components_()
         self.masker_.inverse_transform(weights * self.components_)