How to handle Data, and Images(1) - How to use Numpy

Lesson Notes in .ipynb file

How to handle Data, and Images(1) - How to use Numpy

Topics

• What is Numpy?
• Basic Terms
• Importing Numpy and how to make List/Array with Numpy
• Some basic operations in Numpy
• Merging two array using Numpy
• Using Reshape allows array to be in different shape
• More example on Concatenate and Reshape
• We can also split the array
• Summary

What is Numpy?

• Numpy is a open source library that can be used with Python.
• Since a lot of data can be represented in a matrix format, we can use Numpy to ease this operation.
• Using Numpy is faster than using Python’s List and is a much better solution to a lot of cases.

Basic Terms:

• Vector: 1 dimension list
• Matrix: 2 dimension list
• Tensor: 3 dimension list

Now that we know these terms, we can straight get into how to use Numpy

Importing Numpy and how to make List/Array with Numpy

• import numpy as np: we first need to import the numpy library and to use, we just call ‘np’
• numpy.array(object, dtype=None, *, copy=True, order=’K’, subok=False, ndmin=0, like=None):
  • create a numpy array

import numpy as np

list_data = [1, 2, 3]
arr = np.array(list_data)

print(arr.size)     
print(arr.dtype)    
print(arr[2])       

Output:

3
int64
3

Some basic operations in Numpy

• numpy.arange([start, ]stop, [step, ]dtype=None, *, device=None, like=None):
  • return evenly spaced values within a given interval
• numpy.zeros(shape, dtype=float, order=’C’, *, like=None):
  • returns a new array of given shape and type, filled with zeros
• numpy.ones(shape, dtype=None, order=’C’, *, device=None, like=None):
  • returns a new array of given shape and type, filled with ones
• numpy.random.randint(low, high=None, size=None, dtype=int):
  • returns random integers from low to high
• numpy.random.normal(loc=0.0, scale=1.0, size=None):
  • draw random samples from a normal (Gaussian) distribution

# creating 1d array [0, 1, 2, 3] (from 0 to 3)
# you can think of this like for i in range(4)
arr1 = np.arange(4)
print("arr1:\n", arr1)

# create a 4x4 2d array(matrix) with value float 0
arr2 = np.zeros((4, 4), dtype = float)
print("arr2:\n", arr2)

# create a 3x3 2d array(matrix) with value string 1
arr3 = np.ones((3, 3), dtype=str)
print("arr3:\n", arr3)

# create a 3x3 2d array(matrix) with random values 0 to 9
arr4 = np.random.randint(0, 10, (3, 3))
print("arr4:\n", arr4)

# create a 3x3 2d array(matrix) where mean value is 0 with standard deviation of 1
arr5 = np.random.normal(0, 1, (3, 3))
print("arr5:\n", arr5)

Output:

arr1:
 [0 1 2 3]

arr2:
 [[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]

arr3:
 [['1' '1' '1']
 ['1' '1' '1']
 ['1' '1' '1']]

arr4:
 [[9 2 6]
 [7 9 0]
 [6 5 4]]

arr5:
 [[ 0.66755247 -0.16216641  0.81601341]
 [ 0.7853595  -0.58722737  0.2097311 ]
 [ 1.24401384  0.39303442 -1.37390758]]

Merging two array using Numpy

• numpy.concatenate((a1, a2, …), axis=0, out=None, dtype=None, casting=”same_kind”):
  • join a sequence of arrays along an existing axis

arr1 = np.array([1, 2, 3])
arr2 = np.array([4, 5, 6])
# merge two array arr1, arr2 in a 1D array(vector)
arr3 = np.concatenate([arr1, arr2])

print(arr3.shape)
print(arr3)

Output:

(6,)
[1 2 3 4 5 6]

Using Reshape allows array to be in different shape

• numpy.reshape(a, /, shape=None, order=’C’, *, newshape=None, copy=None):
  • gives a new shape to an array without changing its data

# array 1 is a 1x4 array
arr1 = np.array([1,2,3,4])
# using reshape to make 2x2 array
arr2 = arr1.reshape((2,2))
print("arr2 shape\n", arr2.shape)
print("arr2\n", arr2)

Output:

arr2 shape
 (2, 2)

arr2
 [[1 2]
 [3 4]]

More example on Concatenate and Reshape

# create a 1x4 matrix from 0 to 3
arr1 = np.arange(4).reshape(1, 4)
# create 2x4 matrix from 0 to 7
arr2 = np.arange(8).reshape(2, 4)

print("arr1\n", arr1)
print("arr2\n", arr2)

# merge two array
arr3 = np.concatenate([arr1, arr2], axis=0)
print("arr3\n", arr3)

Output:

arr1
 [[0 1 2 3]]

arr2
 [[0 1 2 3]
 [4 5 6 7]]

arr3
 [[0 1 2 3]
 [0 1 2 3]
 [4 5 6 7]]

We can also split the array

• numpy.split(ary, indices_or_sections, axis=0):
  • split an array into multiple sub-arrays as views into ary.

# create a 2x4 matrix from 0 to 7
arr = np.arange(8).reshape(2, 4)
print("arr\n", arr)

#split the array into 2 vertically
left, right = np.split(arr, [2], axis=1)

print("left\n", left)
print("right\n", right)

Output:

arr
 [[0 1 2 3]
 [4 5 6 7]]
 
left
 [[0 1]
 [4 5]]
 
right
 [[2 3]
 [6 7]]

Summary

• import nupmy as np: importing numpy
• numpy.array: creates a numpy array
• numpy.arange: returns evenly spaced values within a given interval
• numpy.zeros: returns a new array of given shape and type, filled with zeros
• numpy.ones: returns a new array of given shape and type, filled with ones
• numpy.random.randint: returns random integers from low to high
• numpy.random.normal: draw random samples from a normal (Gaussian) distribution
• numpy.concatenate: join a sequence of arrays along an existing axis
• numpy.reshape: gives a new shape to an array without changing its data
• numpy.split: split an array into multiple sub-arrays as views into ary