NumPy is a Python library that can be used to perform advanced mathematical operations. Python has a basic data type called a list. This list data type has limited functionalities. We can perform some basic mathematical operations using these lists, but performing advanced mathematical operations using these lists is really difficult. To address the problem the NumPy library is developed.

The NumPy Python library has a data type called arrays. Using these arrays we can easily implement large and multi-dimensional arrays and matrices. After that, we can use the arrays to perform high-level mathematical functions.

The ancestor of the NumPy library is Numeric. Numeric was first developed by Jim Hugunin with the help of some other developers. Later, Travis Oliphant created the NumPy Python library. Travis Oliphant is also a founding contributor of the SciPy packages in Python.

Please note that the core functionality of the NumPy Python library is its ndarray data structure. These ndarrays are homogeneously typed. In other words, all the elements of an ndarray must be of the same type.

In machine learning, datasets are often stored in a two-dimensional data structure based on ndarrays. The columns of the 2-dimensional data structure store various features of the data and the output. A machine learning model uses these features to predict the output. So, to learn machine learning one should be familiar with the NumPy Python library.

I hope this helps. In the next few articles, we will cover some basic data types of the NumPy library and operations on those data structures.