implemented MAD using scipy (#153)

* implemented MAD using scipy

* refactored _mad using scipy

* added changelog entry

* added changelogs

CITATION.cff

@@ -48,6 +48,9 @@ authors:
     - given-names: Nabil
       family-names: Alibou
       affiliation: 'Biotrial, Neuroscience Department, Rennes, France'
+    - given-names: Ayush
+      family-names: Agarwal
+      affiliation: 'Techno India University, India'
 type: software
 repository-code: 'https://github.com/sappelhoff/pyprep'
 license: MIT

docs/whats_new.rst

@@ -27,6 +27,7 @@ Version 0.5.0 (Unreleased)
 Changelog
 ~~~~~~~~~
 - :meth:`~pyprep.NoisyChannels.find_bad_by_nan_flat` now accepts a ``flat_threshold`` argument, by `Nabil Alibou`_ (:gh:`144`)
+- changed _mad function in utils.py to use median_abs_deviation from the sciPy module, by `Ayush Agarwal`_ (:gh:`153`).
 
 Bug
 ~~~
@@ -234,3 +235,4 @@ Changelog
 .. _Mathieu Scheltienne: https://github.com/mscheltienne
 .. _Ole Bialas: https://github.com/OleBialas
 .. _Nabil Alibou: https://github.com/nabilalibou
+.. _Ayush Agarwal: https://github.com/Ayush-Devs

pyprep/utils.py

@@ -10,6 +10,7 @@
 from psutil import virtual_memory
 from scipy import linalg
 from scipy.signal import firwin, lfilter, lfilter_zi
+from scipy.stats import median_abs_deviation
 
 
 def _union(list1, list2):
@@ -487,11 +488,7 @@ def _mad(x, axis=None):
         raise ValueError(e.format(x.ndim))
 
     # Calculate the median absolute deviation from the median
-    med = np.median(x, axis=axis)
-    if axis == 1:
-        med = med.reshape(-1, 1)  # Transposes array to allow subtraction below
-    mad = np.median(np.abs(x - med), axis=axis)
-
+    mad = median_abs_deviation(x, axis=axis)
     return mad