Fix: ‘numpy.ndarray’ object is not callable in Python

TypeError: 'numpy.ndarray' object is not callable means you put parentheses after a NumPy array, so Python tried to call it like a function. The usual culprits are arr(0) instead of arr[0], or adding () to an attribute like .shape:

arr(0)          # TypeError
arr[0]          # correct
arr.shape()     # TypeError
arr.shape       # correct

Every traceback here is real output from Python 3.12.5, NumPy 2.5.3 and pandas 3.0.6, followed by the fix.

I’ll go through five causes: round brackets for indexing, calling attributes like shape and T, df.values() in pandas, a function replaced by its own result, and how to tell attributes from methods.

Cause 1: Indexing a NumPy array with () instead of []

You read from an array with square brackets; parentheses call something. Swap them and NumPy raises the error right away:

import numpy as np

prices = np.array([19.99, 4.50, 102.00])
print(prices[0])               # indexing uses square brackets
print(prices(0))               # parentheses try to call the array

Output:

19.99
Traceback (most recent call last):
  File "C:\pyguides\ndarray_parens.py", line 5, in <module>
    print(prices(0))               # parentheses try to call the array
          ^^^^^^^^^
TypeError: 'numpy.ndarray' object is not callable
Command Prompt traceback of TypeError numpy.ndarray object is not callable after indexing a NumPy array with parentheses
prices[0] works; prices(0) raises the TypeError.

Plain Python lists give a similar message; see list object is not callable for the list version.

Cause 2: Calling ndarray attributes like shape and T

Several array properties you’ll use are attributes, not methods, so they don’t take parentheses. shape returns a tuple, and calling that tuple fails:

import numpy as np

grid = np.arange(6).reshape(2, 3)
print("shape:", grid.shape)          # attribute: no parentheses
print("size:", grid.size, "| dtype:", grid.dtype)
print("transposed:\n", grid.T)

for name in ["shape", "T"]:
    try:
        getattr(grid, name)()        # what grid.shape() and grid.T() do
    except TypeError as err:
        print(f"grid.{name}() -> TypeError: {err}")

Output:

shape: (2, 3)
size: 6 | dtype: int64
transposed:
 [[0 3]
 [1 4]
 [2 5]]
grid.shape() -> TypeError: 'tuple' object is not callable
grid.T() -> TypeError: 'numpy.ndarray' object is not callable
Command Prompt output of NumPy shape, size, dtype and T attributes, with TypeErrors when shape and T are called with parentheses
grid.shape() fails because shape is a tuple; grid.T() fails because T is already the transposed array.

Note the error message differs: calling shape says “‘tuple’ object is not callable”, while calling T says “‘numpy.ndarray’ object is not callable”.

Cause 3: df.values() in pandas

In pandas, DataFrame.values is an attribute that returns a NumPy array. Adding parentheses calls that array:

import pandas as pd

df = pd.DataFrame({"city": ["Austin", "Denver"], "sales": [1250.50, 980.25]})

print(df.values)                 # .values is an attribute that returns an ndarray
print(df.to_numpy())             # the recommended method form
print(df.values())               # calling it fails

Output:

[['Austin' 1250.5]
 ['Denver' 980.25]]
[['Austin' 1250.5]
 ['Denver' 980.25]]
Traceback (most recent call last):
  File "C:\pyguides\pandas_values.py", line 7, in <module>
    print(df.values())               # calling it fails
          ^^^^^^^^^^^
TypeError: 'numpy.ndarray' object is not callable
Command Prompt traceback of pandas df.values() raising numpy.ndarray object is not callable, after df.values and df.to_numpy() work
df.values and df.to_numpy() both work; df.values() doesn’t.

pandas recommends to_numpy() over .values, and as a method it does take parentheses. For more pandas basics, see loc vs iloc in pandas.

Cause 4: A function was replaced by its NumPy array

Assigning a function’s result back to the function’s name leaves you with an array under that name. The next call fails:

import numpy as np

def monthly_sales():
    return np.array([12500.50, 9800.25, 14300.00])

monthly_sales = monthly_sales()      # the function name now holds the array
print(monthly_sales.sum())
print(monthly_sales())               # calling the array fails

Output:

36600.75
Traceback (most recent call last):
  File "C:\pyguides\ndarray_overwrite.py", line 8, in <module>
    print(monthly_sales())               # calling the array fails
          ^^^^^^^^^^^^^^^
TypeError: 'numpy.ndarray' object is not callable

You can avoid it by giving the result its own name, such as sales = monthly_sales(). In Jupyter, re-run the cell that defines the function, or restart the kernel, to get it back.

How do you know if a NumPy attribute is callable?

I use callable() when I’m unsure. If it returns True, it’s a method and needs (); if False, it’s an attribute and must not have them:

import numpy as np

arr = np.array([3, 1, 2])
for name in ["shape", "T", "size", "sum", "sort", "mean", "dtype"]:
    kind = "method   -> call it ()" if callable(getattr(arr, name)) else "attribute -> no ()"
    print(f"arr.{name:<6} {kind}")

print(arr.sum(), arr.shape)          # method with (), attribute without

Output:

arr.shape  attribute -> no ()
arr.T      attribute -> no ()
arr.size   attribute -> no ()
arr.sum    method   -> call it ()
arr.sort   method   -> call it ()
arr.mean   method   -> call it ()
arr.dtype  attribute -> no ()
6 (3,)
Command Prompt output using Python callable() to show which NumPy array members are methods and which are attributes
shape, T, size and dtype are attributes; sum, sort and mean are methods.

A related trap: naming a variable after a NumPy or built-in function, like sum = np.array(...), then calling sum(...) later. Descriptive names avoid it.

Quick fixes for numpy.ndarray object is not callable

If your code hasChange it to
arr(0)arr[0]
arr.shape()arr.shape
arr.T()arr.T or arr.transpose()
df.values()df.to_numpy()
f = f()result = f()

More NumPy and error-fixing guides:

Frequently asked questions

What does ‘numpy.ndarray’ object is not callable mean?

You used parentheses after a NumPy array, so Python tried to call it like a function.

How do I fix it when indexing?

Use square brackets: arr[0] or arr[1, 2], not arr(0).

Why does arr.shape() raise an error?

shape is an attribute holding a tuple. Use arr.shape without parentheses.

Why does df.values() fail in pandas?

values is an attribute that returns an array. Use df.values or, better, df.to_numpy().

How can I tell if I need parentheses?

Check callable(obj.name). True means it’s a method; False means it’s an attribute.

Where are ndarray attributes documented?

See Array attributes in the NumPy reference.