MSK Institute • Shikohabad
Loading Page Content...
Preparing courses, live labs, and learning resources.सामग्री लोड हो रही है, कृपया प्रतीक्षा करें...
Connecting serverPlease wait...
💻 Practical Labs🚀 Free Demo📜 Verified Certificates
Inspecting Array Attributes (shape, ndim, size, itemsize)0%
Inspecting Array Attributes (shape, ndim, size, itemsize)
Beginner12 min read•Updated: Jul 11, 2026
Inspecting Array Attributes (shape, ndim, size, itemsize)
Every NumPy ndarray encapsulates a lightweight metadata header describing its structure in memory. Being able to inspect an array’s shape, dimensions, memory strides, and byte layout is essential for debugging data pipelines and optimizing neural networks.
1. The Core Four Attributes
Visual Architecture Blueprint
+-----------------------------------------------------------------------------------+ | 1. arr.ndim -> Number of axes / dimensions (e.g., 1 for vector, 2 for matrix)| | 2. arr.shape -> Tuple of integers showing array length along each axis | | 3. arr.size -> Total cumulative count of elements in the entire array | | 4. arr.dtype -> Data type of elements (int32, float64, bool, etc.) | +-----------------------------------------------------------------------------------+
import numpy as np
arr = np.array([
[10.0, 20.0, 30.0, 40.0],
[50.0, 60.0, 70.0, 80.0],
[90.0, 100.0, 110.0, 120.0]
], dtype=np.float64)
print("Dimensions (ndim):", arr.ndim) # 3 rows x 4 cols = 2 dimensions
print("Shape (shape): ", arr.shape) # (3, 4)
print("Total Size (size):", arr.size) # 12 elements (3 * 4)
print("Data Type (dtype):", arr.dtype) # float64
Click to Zoom2. Low-Level Memory Attributes: itemsize, nbytes & strides
For systems programmers and high-performance engineers, NumPy exposes direct hardware layout parameters:
arr.itemsize: Number of bytes consumed by a single element:float64-> 8 bytesint32-> 4 bytesuint8-> 1 bytearr.nbytes: Total bytes consumed by the entire data buffer:
$$ ext{nbytes} = ext{size} imes ext{itemsize}$$ For our (3, 4) float64 array: $12 imes 8 = mathbf{96 ext{ bytes}}$.
arr.strides: The number of bytes the CPU must jump in memory to move 1 step along each axis:- In our (3, 4) float64 array:
arr.stridesis(32, 8). - To advance to the next column (+1 in axis 1), skip 8 bytes (1 float).
- To advance to the next row (+1 in axis 0), skip 32 bytes (4 floats $ imes$ 8 bytes).
3. Practical Array Diagnostics Function
A useful pattern when debugging tensor shapes in data pipelines:
def inspect_array(name, a):
print(f"[{name}] shape={a.shape} | ndim={a.ndim} | size={a.size} | "
f"dtype={a.dtype} | RAM={a.nbytes}B | strides={a.strides}")
# Example:
x = np.ones((100, 200), dtype=np.int32)
inspect_array("FeatureMatrix", x)
# Output: [FeatureMatrix] shape=(100, 200) | ndim=2 | size=20000 | dtype=int32 | RAM=80000B | strides=(800, 4)
Interactive Quiz
Multiple Choice Questions
Question 1
What does the 'shape' attribute return for an array created with 'np.zeros((4, 7, 3))'?
Question 2
If an array has shape (5, 4) and dtype 'int64' (8 bytes), what will 'arr.nbytes' return?
Question 3
What is the relationship between 'arr.size', 'arr.itemsize', and 'arr.nbytes'?
Question 4
What does the 'strides' attribute of an ndarray describe?
Question 5
What will 'np.array(42).ndim' return for a 0-dimensional scalar?
Next Lesson
Identity Matrices & Diagonal Arrays (eye, identity, diag)
Continue learning with hands-on practice, examples, and exercises in the upcoming topic.
Related Lessons
Practice Quiz
Test your understanding of this lesson with 5 questions. Each question has one correct answer.