How to Use argmin and argmax in NumPy for Finding Indexes

Use numpy.argmin() to find the index of the smallest value and numpy.argmax() to find the index of the largest value in a NumPy array. Both functions return the position of these values, which helps locate them quickly in data.

Syntax

numpy.argmin(a, axis=None, out=None) returns the index of the minimum value in the array a. If axis is specified, it returns indices along that axis.

numpy.argmax(a, axis=None, out=None) works similarly but returns the index of the maximum value.

• a: Input array to search.
• axis: Axis along which to find the index. Default is None (flattened array).
• out: Optional output array to store results.

python

numpy.argmin(a, axis=None, out=None)
numpy.argmax(a, axis=None, out=None)

Example

This example shows how to find the index of the smallest and largest values in a 1D and 2D NumPy array.

python

import numpy as np

# 1D array example
arr1d = np.array([4, 2, 7, 1, 9])
min_index_1d = np.argmin(arr1d)
max_index_1d = np.argmax(arr1d)

# 2D array example
arr2d = np.array([[3, 8, 5], [7, 1, 6]])
min_index_2d_axis0 = np.argmin(arr2d, axis=0)  # min indices for each column
max_index_2d_axis1 = np.argmax(arr2d, axis=1)  # max indices for each row

print(f"1D array: {arr1d}")
print(f"Index of min value: {min_index_1d}")
print(f"Index of max value: {max_index_1d}")

print(f"\n2D array:\n{arr2d}")
print(f"Index of min values along axis 0 (columns): {min_index_2d_axis0}")
print(f"Index of max values along axis 1 (rows): {max_index_2d_axis1}")

Output

1D array: [4 2 7 1 9]
Index of min value: 3
Index of max value: 4

2D array:
[[3 8 5]
 [7 1 6]]
Index of min values along axis 0 (columns): [0 1 0]
Index of max values along axis 1 (rows): [1 0]

Common Pitfalls

One common mistake is forgetting that argmin and argmax return the index, not the value itself. To get the value, you must use the index on the array.

Another pitfall is misunderstanding the axis parameter. If you don't specify axis, the array is flattened, and the index is relative to the flattened array, which can be confusing for multi-dimensional arrays.

python

import numpy as np

arr = np.array([[10, 20], [30, 5]])

# Wrong: expecting value, but gets index
min_val_wrong = np.argmin(arr)

# Right: get index, then value
min_index = np.argmin(arr)
min_val_right = arr.flat[min_index]

# Axis confusion example
min_index_axis0 = np.argmin(arr, axis=0)  # min index per column
min_index_axis1 = np.argmin(arr, axis=1)  # min index per row

print(f"Wrong (index instead of value): {min_val_wrong}")
print(f"Right (value using index): {min_val_right}")
print(f"Min indices along axis 0: {min_index_axis0}")
print(f"Min indices along axis 1: {min_index_axis1}")

Output

Wrong (index instead of value): 3
Right (value using index): 5
Min indices along axis 0: [0 1]
Min indices along axis 1: [0 1]

Quick Reference

Remember these quick tips when using argmin and argmax:

• They return the index of the min or max value, not the value itself.
• Use axis to specify the dimension to search along.
• For multi-dimensional arrays, indices correspond to positions along the chosen axis.
• Use the returned index to access the actual value from the array.

Key Takeaways

Use numpy.argmin() and numpy.argmax() to find the index of minimum and maximum values in arrays.

Specify the axis parameter to get indices along a particular dimension in multi-dimensional arrays.

Remember these functions return indices, so use them to access the actual values from the array.

Without axis, the array is flattened and indices refer to the flattened array.

Common mistakes include confusing returned indices with values and misunderstanding axis behavior.