Python Data Analysis - Second Edition

Creating a NumPy masked array

Data is often messy and contains gaps or characters that we do not deal with often. Masked arrays can be utilized to disregard absent or invalid data points. A masked array from the numpy.ma subpackage is a subclass of ndarray with a mask. In this section, we will use the face photo as the data source and act as if some of this data is corrupt. The following is the full code for the masked array example from the ch-04.ipynb file in this book's code bundle:

import numpy import scipy import matplotlib.pyplot as plt face = scipy.misc.face() random_mask = numpy.random.randint(0, 2, size=face.shape) plt.subplot(221) plt.title("Original") plt.imshow(face) plt.axis('off') masked_array = numpy.ma.array(face, mask=random_mask) ...