Easy NumPy Tutorial Part 3

Data Science May 29, 2021

Introduction

The Part 3 of easy numpy tutorial series comprises of Sorting, Set Operations, File I/O and Linear Algebra.

Sorting

ndarrays can be sorted with sort method:

arr = np.random.randn(10)
arr.sort()
# Array is now sorted in ascending order

For multidimensional arrays you can pass the axis which you want to be sorted:

arr = np.random.randn(3, 4)
arr.sort(axis=1)  # Sort each row

The module level np.sort returns a new sorted copy of the array instead of modifying it.

Some Set Operations

NumPy provides some basic set operations for one dimensional arrays. The most common is unique which returns sorted unique values:

arr = np.array(['apple', 'mango', 'apple', 'banana', 'avocado', 'mango'])
np.unique(arr)
# Output: array(['apple', 'avocado', 'banana', 'mango'], dtype='<U7')

File Input/Output with NumPy Arrays

NumPy provides an easy to use API for loading and saving data from and to disks in text or binary format.

np.save function saves the data in raw binary with .npy extension:

arr = np.arange(10)
np.save('array', arr)

The file on the disk can be loaded with np.load:

np.load('array.npy')
# Output: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

Linear Algebra with NumPy

Common linear algebra operations like matrix multiplication, determinant can easily be done with NumPy.

You can perform array dot product with dot:

x = np.array([[1, 2, 3, 4], [5, 6, 7, 8]])
y = np.array([[9, 10], [11, 12], [13, 14], [15, 16]])

x.dot(y)
# Output: array([[130, 140],
#                [322, 348]])

As of Python 3.5, you can also use @ for matrix multiplication:

x @ y
# Same result as x.dot(y)

numpy.linalg provides standard functions like matrix decomposition, inverse, determinant etc:

from numpy.linalg import inv

arr = np.random.randn(2, 2)
inv(arr)  # Returns the inverse of the matrix

Conclusion

After completing this tutorial, you must be familiar with important concepts of NumPy, which you will find useful in various numerical computing and data analysis problems.

Check out Part 1 and Part 2 if you haven’t already!

References

Some examples and topics are referenced from Python for Data Analysis: Data Wrangling with Pandas, NumPy, and IPython by Wes McKinney. It is great for beginners who want to learn data analysis with python.