find_outliers¶

find_outliers(data: ~numpy.ndarray, outliers: float, deviation: callable = <function std>, center: callable = <function mean>) → ndarray[bool][source][source]¶

Find outliers in data matrix.

This function finds outliers in a data matrix. Outliers are defined as any trial with a maximum value greater than the mean plus outliers times the standard deviation. The function returns a boolean array with True for trials that are not outliers and False for trials that are outliers.

Parameters:

data (np.ndarray) – Data to find outliers in. (trials X channels X (frequency) X time)
outliers (float) – Number of deviations from the mean to consider an outlier.
deviation (callable, optional) – Metric function to determine the deviation from the center. Default is median absolute deviation.
center (callable, optional) – Metric function to determine the center of the data. Default is median.

Returns:

Boolean array with True for trials that are not outliers and False for trials that are outliers.

Return type:

np.ndarray[bool]

Examples

>>> import numpy as np
>>> data = np.array([[1, 1, 1, 1, 1], [0, 60, 0, 10, 0]]).T
>>> find_outliers(data, 1)
array([ True, False,  True,  True,  True])
>>> find_outliers(data, 10)
array([ True,  True,  True,  True,  True])
>>> find_outliers(data, 0.1)
array([ True, False,  True, False,  True])
>>> find_outliers(data, 0.1, np.std)
array([ True, False,  True, False,  True])
>>> data = np.array([[1, 1, np.nan, 1, 1], [0, 60, 0, 10, 0]]).T
>>> find_outliers(data, 0.1, st.median_abs_deviation)
array([ True, False,  True, False,  True])