Recently, I was working on a data analysis project where I needed to reset the index of a NumPy array after filtering out some values. The issue is, there’s no direct “reset_index()” method in NumPy like we have in pandas. So we need a workaround.
In this article, I’ll cover several simple ways you can use to reset or reindex NumPy arrays in Python (including basic indexing techniques and more advanced approaches).
So let’s get in!
NumPy Reset Index
Now, I will explain to you the methods to reset the index in Python NumPy.
Read np.genfromtxt() Function in Python
Method 1 – Use Simple Slicing
NumPy arrays are indexed differently from pandas DataFrames. When you filter a Python NumPy array, it creates a view or copy with only the selected elements, but doesn’t maintain any concept of the original indices.
Let’s see this with an example:
import numpy as np
# Create a sample array
data = np.array([10, 20, 30, 40, 50, 60, 70, 80, 90])
# Filter values greater than 50
filtered_array = data[data > 50]
print("Filtered array:", filtered_array)Output:
Filtered array: [60 70 80 90]You can see the output in the screenshot below.
The filtered array now contains only the values greater than 50, but there’s no built-in way to know these came from indices 5, 6, 7, and 8 of the original array. If you need to maintain this relationship, here’s how you can do it:
import numpy as np
# Create a sample array
data = np.array([10, 20, 30, 40, 50, 60, 70, 80, 90])
# Get both indices and values where condition is met
indices = np.where(data > 50)[0]
values = data[indices]
print("Original indices:", indices)
print("Values:", values)
# If you need a "reset" index (0, 1, 2, 3...)
new_indices = np.arange(len(values))
print("New indices:", new_indices)Output:
Original indices: [5 6 7 8]
Values: [60 70 80 90]
New indices: [0 1 2 3]This approach gives you both the original indices and the values, along with a new sequential index starting from 0.
Check out np.savetxt() Function in Python
Method 2 – Use numpy.nonzero()
Another approach is to use np.nonzero() which returns the indices of non-zero elements in the array in Python. We can combine this with a boolean mask to get indices where our condition is true:
import numpy as np
# Create a sample array
data = np.array([10, 20, 30, 40, 50, 60, 70, 80, 90])
# Get indices of elements that satisfy the condition
condition = data > 50
indices = np.nonzero(condition)[0]
values = data[indices]
print("Original indices:", indices)
print("Values:", values)
# Reset indices to start from 0
result = np.column_stack((np.arange(len(indices)), values))
print("Array with reset indices:\n", result)Output:
Original indices: [5 6 7 8]
Values: [60 70 80 90]
Array with reset indices:
[[ 0 60]
[ 1 70]
[ 2 80]
[ 3 90]]You can see the output in the screenshot below.
This gives us a 2D array where the first column contains the new indices (0, 1, 2, 3) and the second column contains our values.
Read NumPy Array to List in Python
Method 3 – Create a Structured Array
If you want something more similar to pandas’ behavior, you can create a structured array that keeps track of both indices and values in Python:
import numpy as np
# Create a sample array
data = np.array([10, 20, 30, 40, 50, 60, 70, 80, 90])
# Filter values
mask = data > 50
filtered_values = data[mask]
# Create a structured array with original index and value
original_indices = np.where(mask)[0]
structured_array = np.zeros(len(filtered_values), dtype=[('original_index', int), ('value', int)])
structured_array['original_index'] = original_indices
structured_array['value'] = filtered_values
print("Structured array with original indices:")
print(structured_array)
# "Reset" the index
reset_structured_array = np.zeros(len(filtered_values), dtype=[('new_index', int), ('original_index', int), ('value', int)])
reset_structured_array['new_index'] = np.arange(len(filtered_values))
reset_structured_array['original_index'] = original_indices
reset_structured_array['value'] = filtered_values
print("\nStructured array with reset indices:")
print(reset_structured_array)Output:
Structured array with original indices:
[(5, 60) (6, 70) (7, 80) (8, 90)]
Structured array with reset indices:
[(0, 5, 60) (1, 6, 70) (2, 7, 80) (3, 8, 90)]You can see the output in the screenshot below.
This method gives you a structured array that maintains both the original indices and the new reset indices, along with the values.
Check out NumPy Reverse Array in Python
Method 4 – Use a Custom Function
For convenience, we can create a custom function to handle the index resetting:
import numpy as np
def reset_index(arr, condition=None):
"""
Reset index of a NumPy array, optionally filtering by a condition.
Returns a structured array with new_index, original_index, and value.
"""
if condition is not None:
original_indices = np.where(condition)[0]
values = arr[condition]
else:
original_indices = np.arange(len(arr))
values = arr
result = np.zeros(len(values), dtype=[('new_index', int), ('original_index', int), ('value', int)])
result['new_index'] = np.arange(len(values))
result['original_index'] = original_indices
result['value'] = values
return result
# Usage example
data = np.array([10, 20, 30, 40, 50, 60, 70, 80, 90])
result = reset_index(data, data > 50)
print(result)Output:
[(0, 5, 60) (1, 6, 70) (2, 7, 80) (3, 8, 90)]This function gives you a clean way to reset the index of any NumPy array while preserving the original indices if needed.
Read NumPy Array to a String in Python
Method 5 – Use Pandas to Reset Index
If you’re working in a data analysis context where you’re likely already using pandas, you can leverage pandas’ reset_index method:
import numpy as np
import pandas as pd
# Create a sample array
data = np.array([10, 20, 30, 40, 50, 60, 70, 80, 90])
# Convert to pandas Series, filter, and reset index
series = pd.Series(data)
filtered_series = series[series > 50].reset_index()
filtered_series.columns = ['original_index', 'value']
# Add a new reset index
filtered_series['new_index'] = np.arange(len(filtered_series))
print(filtered_series)
# Convert back to NumPy if needed
result_array = filtered_series.to_numpy()
print("\nAs NumPy array:")
print(result_array)Output:
original_index value new_index
0 5 60 0
1 6 70 1
2 7 80 2
3 8 90 3
As NumPy array:
[[5 60 0]
[6 70 1]
[7 80 2]
[8 90 3]]While this approach involves converting between NumPy and pandas, it can be very convenient if you already use pandas in your workflow.
I hope you found this article helpful. The key takeaway is that NumPy doesn’t have a direct “reset_index” function like pandas, but there are several ways to achieve the same result depending on your needs. Whether you want to create a new sequential index, keep track of the original indices, or create a more structured representation, the above methods should cover most use cases.
You may read:
- np.diff() Function in Python
- Replace Values in NumPy Array by Index in Python
- np.add.at() Function in Python
I am Bijay Kumar, a Microsoft MVP in SharePoint. Apart from SharePoint, I started working on Python, Machine learning, and artificial intelligence for the last 5 years. During this time I got expertise in various Python libraries also like Tkinter, Pandas, NumPy, Turtle, Django, Matplotlib, Tensorflow, Scipy, Scikit-Learn, etc… for various clients in the United States, Canada, the United Kingdom, Australia, New Zealand, etc. Check out my profile.
