NumPy is one of the most popular packages in the Python ecosystem. NumPy adds support to large multidimensional arrays and matrices with great efficiency.

Numpy arrays are a lot faster than traditional python lists because they are stored in continuous memory that allows faster processing and accessing thus making large data processing a lot faster and convenient.

In this tutorial, we will learn how to create NumPy arrays.

## Prerequisite

You must have Python3+ installed in your system, if that’s not the case following articles can help you.

- How To Install Python 3 on Mac OS
- How To Install Python On Windows
- How To Create Python Virtual Environments On Ubuntu

Second, you must have the latest version of NumPy installed in your system. Use the following command to install NumPy.

```
pip install numpy
```

## How to create NumPy arrays

There are multiple ways you can create an array in Numpy, let’s go over the most common ones one by one.

## Creating NumPy array with arrange function

NumPy comes with a built-in method `arrange()`

that’s quite similar to the range() function in Python.

```
import numpy as np
a = np.arange(11) # creates a range from 0 to 10
print(a)
print(a.shape)
```

**Output**

```
[ 0 1 2 3 4 5 6 7 8 9 10]
(11,)
```

First, we import the NumPy package as `np`

then we invoked the `arrange()`

method with a value. Since we passed only one value it will create a rank 1 array or one-dimensional array of 11 elements.

Noticed how indexing starts from 0? Therefore we don’t have the number 11 on the list.

The `shape`

property returns the shape of the array which is a one-dimensional array with 11 elements or you can think of it as an 11*1 matrix.

Note– Arrays and Matrix are often used interchangeably. It’s very crucial to know the difference between the two in terms of NumPy.Numpy

matricesare strictly 2-dimensional, while NumPyarrays(ndarrays) are N-dimensional. Matrix objects are a subclass of`ndarray`

, so they inherit all the attributes and methods of`ndarrays`

.As per the official documents, it’s not advisable to use matrix class since it will be removed in the future – https://numpy.org/doc/stable/reference/generated/numpy.matrix.html

You can play around with the `arrange()`

function by passing negatives values such as.

```
import numpy as np
a = np.arange(-5, -1)
print(a)
```

**Output**

```
[-5 -4 -3 -2]
```

As you can observe the default step between the elements is 1, this can also be configured.

```
import numpy as np
a = np.arange(0,11,2) # creates a range from 0 to 10, step 2
print(a)
```

**Output**

```
[ 0 2 4 6 8 10]
```

Similar to python’s `range()`

function we can pass attributes such as start, stop and step to the `arrange()`

function.

`arange(start, stop, step)`

This step function can be used to count backward as well.

```
import numpy as np
a = np.arange(5, 1, -1)
print(a)
```

**Output**

```
[5 4 3 2]
```

## Creating NumPy array with zeros function

```
import numpy as np
a = np.zeros(5) # create an array with all 0s
print(a) # [ 0. 0. 0. 0. 0.]
print(a.shape)
```

**Output**

```
[0. 0. 0. 0. 0.]
(5,)
```

Now let’s create some two-dimensional arrays with the `zeros()`

method.

```
import numpy as np
a = np.zeros((2,3)) # array of rank 2 with all 0s; 2 rows and 3 columns
print(a.shape)
print(a)
```

**Output**

```
(2, 3)
[[0. 0. 0.]
[0. 0. 0.]]
```

The `zeros()`

method takes `shape`

as the first argument

`zeros(shape)`

This shape could be an int for a 1D array and a tuple of ints for an N-D array. Thus we passed a tuple as an argument to create a 2D array where the former is the number of rows and the latter is the number of columns.

`np.zeros((2,3)) # array of rank 2 with all 0s; 2 rows and 3`

Let’s go ahead and create a 3D array with zeros.

```
import numpy as np
a = np.zeros((3,2,3)) # array of rank 3 with all 0s; 2 rows and 3
print(a.shape)
print(a)
```

**Output**

```
(3, 2, 3)
[[[0. 0. 0.]
[0. 0. 0.]]
[[0. 0. 0.]
[0. 0. 0.]]
[[0. 0. 0.]
[0. 0. 0.]]]
```

To create a 3D array and above we need to pass a tuple with 3 elements to the `zeros()`

method. Here the first element is the rank of the array and the latter are the number of rows and column respectively.

Other than shape zeros take two more arguments.

`numpy.zeros(shape, dtype=float, order='C')`

- Here
`dtype`

is the datatype of the elements which by default is float. - The order parameter determines how the data is stored in memory.
**C**: C-style of storing multi-dimensional data in row-major memory.**F**: Fortran style of storing multi-dimensional data in column-major in memory.

So in case you want to create a zeros array with only integers you would wanna do something like this.

```
import numpy as np
a = np.zeros((2,3), int) # array of rank 2 with all 0s; 2 rows and 3 with integer values
print(a)
```

**Output**

```
[[0 0 0]
[0 0 0]]
```

## Creating Numpy arrays with the full function

Instead of 0’s let’s say you want to creates arrays full of a different number for instance let the number be 6, then you could use the `full()`

function of NumPy.

```
import numpy as np
a = np.full((2,3), 6) # array of rank 2 with all 6s
print(a)
```

**Output**

```
[[6 6 6]
[6 6 6]]
```

The function is quite similar to the `zeros()`

function which we saw earlier this one just takes an additional argument `fill_value`

.

`np.full(shape, fill_value, dtype = None, order = 'C')`

## Creating NumPy arrays with eye function

Sometimes, you need to create an array that mirrors an identity matrix. You can do so using the `eye()`

function.

For those who don’t remember what an identity matrix is – A square matrix in which all the elements of the principal diagonal are ones and all other elements are zeros.

`a = np.eye(4) # 4x4 identity matrix`

**Output**

```
[[1. 0. 0. 0.]
[0. 1. 0. 0.]
[0. 0. 1. 0.]
[0. 0. 0. 1.]]
```

## Creating NumPy array of random numbers

```
import numpy as np
a = np.random.random((2,4)) # rank 2 array (2 rows 4 columns) with random values in the half-open interval [0.0, 1.0)
print(a)
```

**Output**

```
[[0.60750379 0.86313917 0.06771915 0.38928062]
[0.03435306 0.45732639 0.50627336 0.69863683]]
```

Similarly, you can create a 3D array using the `random()`

function.

`a = np.random.random((3,2,4))`

**Output**

```
[[[0.98981043 0.87668365 0.60244216 0.18769144]
[0.95372264 0.92507173 0.74217822 0.39551042]]
[[0.03540292 0.43823864 0.28823769 0.71393248]
[0.84661176 0.58682121 0.33108038 0.91641688]]
[[0.83330768 0.40824907 0.01822309 0.2082025 ]
[0.77093124 0.72571395 0.81175314 0.08715285]]]
```

Notice how all the values are within the range of (0,1)?

What if you want to create a uniform matrix within a different range?

For example a 2D array within the range of 1 to 10 having 2 rows and 3 columns. You can do so using the `random.uniform()`

function.

```
import numpy as np
a = np.random.uniform(low=1, high=10, size=(2,3))
print(a)
```

**Output**

```
[[6.14970466 9.91562178 6.58209242]
[4.83473852 3.28020197 3.06821119]]
```

## Creating NumPy array from a Python list

Of course, python allows the creation of NumPy array from a regular python list.

```
import numpy as np
some_list = [1,2,3,4,5]
arr = np.array(some_list)
print(arr)
```

**Output**

`[1 2 3 4 5]`