*Numpy* package is one the most widely used library in the Python environment. It enables straightforwardly vectorising the operations. In the world of Deep Learning and Neural Networks, this is a desirable quality.

Numpy's main object is a multidimensional array storing elements of the same type, usually numbers. The arrays (*umpy.array*) created using the library differ from the ones in Python standard library (*array.array*). Let us first run the Python command line and import the package.

`python`

`import numpy as np`

There are several ways on how to create the arrays. The easiest is to provide the set of value, by passing a regular Python list or tuple using the array function.

```
np.array([2, 4, 15])
np.array([[1, 2], [3, 4], [5, 6]])
```

The transformations change sequences of sequences into two-dimensional arrays, sequences of sequences of sequences into three-dimensional arrays, and so on. Notice that the following line will return with the error:

`np.array(2, 4, 15)`

To find out about the shape of the array use the following:

```
a = np.array([[1, 2], [3, 4], [5, 6]])
a.shape
```

To change it, use the reshape function:

`a.reshape(2, 3)`

You may find the functions zeros and ones useful to create arrays containing only 0s or 1s:

```
np.zeros((3, 4))
np.ones((2, 3))
```

For the randomly generated numbers apply:

`np.random.random((2,3))`