NumPy

Updated Aug 17, 2021 ·

Overview

Python lists are flexible but slow for data analysis, especially when working with large datasets. NumPy (Numeric Python) solves this by providing NumPy arrays, which enable fast, element-wise calculations.

To use NumPy:

• Install it.

  pip install numpy 

• Import it using:

  import numpy as np

Usage

You can create a NumPy array with np.array(). Unlike lists, NumPy arrays let you perform operations like BMI calculation across all elements efficiently, making it a powerful tool for data analysis.

Consider the two NumPy arrays below for height and weight for a sports team comprising of 6 athletes.

import numpy as np 

height = [1.56, 1.62, 1.75, 1.80, 1.85, 1.89]  # Heights in meters
weight = [65.3, 68.5, 72.4, 76.0, 80.2, 89.1]  # Weights in kilograms

np_height = np.array(height)
np_weight = np.array(weight)

If you try to calculate the BMI directly with lists, Python will raise an error because mathematical operations like squaring (**) or division cannot be applied element-wise to lists:

bmi = weight / height ** 2
print(bmi)

Output:

TypeError: unsupported operand type(s) for ** or pow(): 'list' and 'int' 

Using NumPy arrays, you can easily perform the BMI calculation:

bmi = np_weight / np_height ** 2 
print(bmi)

Output:

array([26.83267587, 26.10120408, 23.64081633, 23.45679012, 23.43316289, 24.94331066]) 

NumPy Array Type Rules

NumPy arrays are designed to hold values of a single type, such as floats, booleans, or strings. It assumes that the values inside an array are of the same type.

• If you mixed types, they are automatically converted to one type
• A NumPy array is simply a new kind of Python type.
• it comes with unique methods that behave differently from lists.

For example, the array below contains different types.

np.array([1.0, "is", True])  

The boolean and the float will be converted to strings, as shown in the output:

array(['1.0', 'is', 'True'], dtype='<U32')

Python Lists vs. NumPy Arrays

Operations on Python lists and NumPy arrays can produce different results.

• Adding two Python lists combines their elements into a single list.

  python_list = [1, 2, 3] 
  python_list

  Output:

  [1, 2, 3, 1, 2, 3]

• Adding two NumPy arrays performs element-wise addition.

  numpy_array = np.array([1, 2, 3])
  numpy_array + numpy_array

  Output:

  array([2, 4, 6])

NumPy Subsetting

Subsetting works similarly to lists using square brackets. NumPy supports boolean subsetting:

• Create a boolean array by comparing values.
• Use the boolean array for subsetting.
• This filters data and provides useful insights.

Using the previous example:

import numpy as np 

height = [1.56, 1.62, 1.75, 1.80, 1.85, 1.89]  # Heights in meters
weight = [65.3, 68.5, 72.4, 76.0, 80.2, 89.1]  # Weights in kilograms

np_height = np.array(height)
np_weight = np.array(weight)

bmi = np_weight / np_height ** 2 
print(bmi)

Output:

array([26.83267587, 26.10120408, 23.64081633, 23.45679012, 23.43316289, 24.94331066]) 

To retrieve the BMI of the third athlete, use an index (remember, indexing starts at 0):

bmi[2]

Output:

23.640816326530615

To filter BMIs greater than 24, first create a boolean array by performing the comparison:

bmi > 24 

This will return a Numpy array of boolean values.

array([ True,  True, False, False, False,  True])

You can then use this boolean array to subset the original BMI array:

bmi[bmi > 24]

Output:

array([26.83267587, 26.10120408, 24.94331066]) 

Comparison and Boolean Operators

Operational operators like < and >= worked with NumPy arrays out of the box. They allow you to perform element-wise comparisons without additional modifications.

Example:

import numpy as np
my_array = np.array([10, 15, 20, 25])
print(my_array > 15)  

Output:

[False False  True  True]  

Unfortunately, this is not true for the boolean operators like and, or, and not. To use these operators with NumPy, you will need to use the equivalent:

• np.logical_and()
• np.logical_or()
• np.logical_not()

Consider two NumPy arrays representing house areas:

import numpy as np
house_x = np.array([18.0, 20.0, 10.75, 9.50])
huose_y = np.array([14.0, 24.0, 14.25, 9.0]) 

To determine the areas in house_x which greater than 18.5 but smaller than 10:

print(np.logical_or(house_x > 18.5, house_x < 10))

Output:

[False  True False  True]

To find areas smaller than 11 in both house_x and huose_y:

print(np.logical_and(house_x < 11, huose_y < 11))

Output:

[False False False  True]