Counting Elements in Multidimensional Arrays with Python: Simple Tricks

Counting Elements in Multidimensional Arrays

Problem

While counting elements in a one‑dimensional array is straightforward, counting each element in a multidimensional array requires additional techniques.

Thought Process

Common Python libraries do not provide a direct, simple method for multidimensional arrays, so the problem can be solved indirectly by flattening the array first and then counting.

Two‑Dimensional Array

Data preparation using NumPy with a fixed random seed:

import numpy as np
import pandas as pd

np.random.seed(2022)  # set random seed
a = np.random.randint(0, 10, (100, 3))  # 100×3 2D array

Counting elements with collections.Counter after flattening:

from collections import Counter

# Counter cannot directly handle a NumPy array; flatten it first
Counter(a.flatten())
# Or use the flat iterator
Counter(a.flat)

Using pandas value_counts after melting the DataFrame:

# Melt the DataFrame to a single column named 'value' and count
pd.DataFrame(a).melt().value_counts('value')

Multidimensional Array

Regular (uniform) 3‑D array generation:

np.random.seed(2022)
a = np.random.randint(0, 10, (10, 3, 4))  # 10×3×4 3D array

For uniform arrays, .flatten() easily converts them to one dimension, after which counting proceeds as with 2‑D arrays.

Pandas cannot directly convert higher‑dimensional arrays to its DataFrame structure, so the flatten‑then‑count approach remains preferable.

Irregular (jagged) Array

When the array consists of nested lists of varying lengths, numpy.flatten() is insufficient. Example of an irregular nested list:

list_b = [[[1,2,3,4,[1,2,3,4,[1,2,3,4]]],6,7],8,9,[1,2,3,4,[1,2,3,4]],6,7]]
np.array(list_b, dtype='O').flatten()

Because the flattened result contains objects, Counter cannot directly count them. A custom recursive flatten function can handle such structures:

def flatten(values):
    # Convert to a NumPy flat iterator first
    values = np.array(values, dtype='O').flat
    for value in values:
        if isinstance(value, (list, np.ndarray)):
            yield from flatten(value)
        else:
            yield value

Using this function, the irregular array can be counted via pandas:

pd.DataFrame(list_b).stack().map(flatten).map(tuple).explode().value_counts()

The combination of stack (or melt) and the custom flatten function produces a pandas Series that can be counted with value_counts.

Summary

The article demonstrates several Python‑based methods—NumPy flattening, collections.Counter, and pandas melt/value_counts —to count element frequencies in both regular and irregular multidimensional arrays, and provides a reusable recursive flatten function for complex nested structures.