Although it may seem unlikely to a beginner, plotting is a critical part of a data science job. You need to master visualization, not only to communicate the results of your work, or make presentations, but to understand the problem itself just at the beginning of your work on a dataset. This phase is called explorative analysis and helps you define the problem structure in the most suitable terms. You can see that plotting tools are the eyes and tongue of data science.
Visualization is also an art. To make the best graphics for your project, you should carefully choose the plotting libraries that suit your artistic style and your understanding of the problem; and master their tricks and capabilities. The multitude of Python visualization libraries makes finding the right one more challenging. We have gathered and compared all major Python visualization libraries here so you can choose easier.

Matplotlib

Documentation: matplotlib.org
Matplotlib is the first and most popular option. This library is an essential part of any python data science environment. Many programmers initially ignore it and go for simpler, more stylish options out there. But if you take your time and learn the tools and tricks of this library, you will be surprised by its power and complete artillery of visualization tools. You can build nearly any diagram using Matplotlib,and its somewhat unattractive default style has become more colorful in the newer versions.
Matplotlib can be used in nearly any environment. Including Python scripts, the Python and IPython shells, the Jupyter notebook, web application servers and four graphical user interface toolkits.
As this library is older than the HTML5 canvas, it produces its figures in the form of static images. Matplotlib can also generate interactive figures using desktop-GUI toolkits like Qt and GTK.
This popular library has several caveats as everything else. Matplotlib code for making a simple chart is generally longer than other libraries, having a non-pythonic structure because the library was originally designed for Matlab. It’s dual interfaces for functional and object-oriented code makes it more difficult to find help online.

Pandas

Documentation: pandas.pydata.org
This well-known Python data analysis library includes a built-in plotting capability which enables the user to draw the basic charts and diagrams quickly. These methods are added to make exploratory data analysis faster, thus they do not give you fancy, customizable, publication-ready charts. If you want to customize your graphs, you must learn matplotlib.

Seaborn

Documentation: seaborn.pydata.org
Seaborn is not a standalone visualization library, but a thin layer Based on matplotlib. You need to have the old package installed to be able to use the new one.
Seaborn adds a simpler interface to matplotlib. It provides color scheme choices and the default styles are more beautiful. These features make plotting and styling complex charts much simpler.
Seaborn works well with pandas. Although It supports some of the more complex visualization approaches, you still need to know some matplotlib to give some nice touches to your diagrams.

Ggplot.

Documentation: yhat.github.io
Python Ggplot library is a port of ggplot2 for R, and makes charts based on the grammar of graphics (search it if you’re curious) as its R version. Its intuitive grammar makes ggplot simple, powerful and at the same time less customizable than matplotlib. If you are used to matplotlib, the R-like API may be a little hard to grasp because of the non-pythonic nature of ggplot.
Ggplot lets you add components layer by layer -like in photoshop-. Meaning you can first draw the axis, then the points, the tags, some extra markings,…. In different stages of your program.

Bokeh.

Documentation: docs.bokeh.org
Bokeh is a very powerful, native python library suitable for generating browser-ready, interactive diagrams from large, real-time or streaming data. This makes using Bokeh for your small home project a little overkill. But ideal for your medium-to-large website’s real-time statistics dashboard.
Bokeh follows the grammar of graphics, and generates visualizations that are later rendered for modern browsers. These visualizations may be in the form of JSON objects, standalone HTML documents, or interactive web applications.
Bokeh can generate a large variety of aesthetically pleasing charts in Python, R, Scala and some other environments. It has a three-level interface based on the level of complexity and customizability. The highest level is the fastest and simplest one; and the lowest one is the most customizable, suitable for complex plotting by experienced programmers.