Question

I have a numpy array “data” that just contains a set of integer counts. Given another array “bins”, I just want to make a frequency plot/CDF of the fraction of total entries in “data” that have at least bins[0]-many counts, at least bins[1]-many counts, etc. and make it into a bar plot, in matplotlib. For example, if:

data = [1, 4, 5, 10]
bins = [0, 5, 6, 7]

then the result should be a bar graph that has 0, 5, 6, 7 on the x-axis and then the fraction of data that has values >= 0, then values >= 5, etc. How can I make this kind of “discrete” CDF bar plot with specified bins in matplotlib? Thanks.

Answer 1

If you’re using matplotlib I assume you are also using numpy, so you can just go through bins and work out the fraction of data such that data>bin, for all data in datas and for all bin in bins.

To that effect this could work:

import numpy as np
# turn data into numpy array for easier manipulation
data2 = np.array(data)
n     = len(data2)

# calculate fractions for each bin in bins
# astype('float') because otherwise you end up doing integer arithmetic
fracs = [ sum(data2>=bin).astype('float')/n for bin in bins ]

Now just plot bins against fracs, e.g.

import matplotlib.pyplot as plt
plt.bar(bins,fracs)
plt.show()