Efficiently Determining Frequency Counts of Unique Values in NumPy Arrays

This article explores an efficient method for calculating the frequency counts of unique values within a NumPy array.

Using numpy.unique with return_counts=True (for NumPy versions 1.9 and above) allows for efficient computation of both unique values and their corresponding counts. For illustration:

import numpy as np

x = np.array([1,1,1,2,2,2,5,25,1,1])
unique, counts = np.unique(x, return_counts=True)

print(np.asarray((unique, counts)).T)

This approach significantly outperforms the scipy.stats.itemfreq function in terms of execution speed, as demonstrated in performance benchmarks:

In [4]: x = np.random.random_integers(0,100,1e6)

In [5]: %timeit unique, counts = np.unique(x, return_counts=True)
10 loops, best of 3: 31.5 ms per loop

In [6]: %timeit scipy.stats.itemfreq(x)
10 loops, best of 3: 170 ms per loop