6.0 Introduction

This lesson introduces basics of plotting in python.

Pure python does not contain a built-in module for plotting. Fortunately, other people have developed many awesome tools for plotting in python. However, since these tools do not come together with python, it means we have to install them. One such tool is matplotlib. Once the matplotlib is installed, we can import it as below:

import matplotlib

the pyplot sub-module of matplotlib as

import matplotlib.pyplot as plt

It was developed to replicate MATLAB in python. It is a very comprehensive and probably the most widely used plotting library in python. matplotlib provides great flexibility to the user to accomplish data visualization in python. However, this flexibility comes with a price. Suppose if you want to build a house, you have a choice either to make walls, doors windows, etc yourself or you can get these ready-to-use materials and then join them together. Using the first approach is more cumbersome and time consuming because you will only have raw material (wood, rocks, soil, water) to start with. However, it provides you the flexibility which you will not have if you use the second approach. The same is true for matplotlib. There are two APIs in matplotlib, the functional and object oriented. Following example shows drawing a plot using object oriented API of matplotlib.

fig, axes = plt.subplots()
axes.plot([1,2,3])
plt.show()

If we want to add the legend, we have to add another line as below

_, axes = plt.subplots()
axes.plot([1,2,3], label="a line")
plt.legend()
plt.show()

Above we first cretae the axes and then use it plot the list of values. We have to call plt.show() to see the plot. Once the plot is shown, the axes is exhausted. This means we can not work on it any further. We have to create a new axes if we want to plot a new line. Above we have created the plot using objected oriented API of matplotlib. There is also a functional API in matplotlib. To accomplish above using functional API, we can do as below

plt.plot([1,2,3], label="a line")
plt.legend()
plt.show()

As we can notice that, the functional API will fall short to our requirements most of the times. But the objected oriented API seems too verbose. What if we could do

plot([1,2,3], label="a line")

and we would be able to see the plot. This necessity gave rise to easy_mpl library. The purpose of easy_mpl is to ease the use of matplotlib while keeping the flexibility of object oriented programming paradigm of matplotlib intact. Using these one liners will save the time and will not hurt. Moreover, you can swap most function of this library with that of matplotlib and vice versa. In this chapter, we will mainly use easy_mpl library for plotting.

easy_mpl contains two kinds of functions, one which are just wrappers around their matplotlib alternatives. These include plot(), scatter(), bar_chart(), pie(), hist(), imshow() and boxplot(). As the name suggests, these are just alternatives to their matplotlib aliases. All of these functions take same input arguments as taken by corresponding matplotlib functions. If these functions are given same input arguments as to their matplotlib alternatives, then these functions return same output as returned by matplotlib. Therefore, we can consider them as alternative to matplotlib (for most cases). All these functions take three further input arguments. These are ax, ax_kws and show. The meanings of these three arguments are as below

• ax stands for axes, the matplotlib axes object matplotlib.axes. If ax argument is given, then the plots are drawn on this otherwise either a new matplotlib axes is created or currently available axes is used.

• ax_kws is a dictionary which includes the arguments to manipulate the x and y labels, ticklabels, title. These arguments are passed to easy_mpl.utils.process_axes() function.

• The show argument determines whether to draw the plot after the function or not. If show is set to False, then the axes is not exhausted, which means, we can manipulate it if required and call plt.show or plt.draw after manipulating the axes. Otherwise, in default case (when show is True), the plot is drawn immediately after calling the corresponding function.

Moreover these wrapper functions also take some auxiliary input arguments which can be used for further manipulation of these plots. For example the imshow() function takes the grid_params argument. The second kinds of functions in this library are helper functions for data visualization and analysis. These include regplot(), dumbbell_plot(), ridge(), parallel_coordinates(), taylor_plot(), lollipop_plot(), circular_bar_plot(), violin_plot() and spider_plot() . Thus easy_mpl is not a replacement to matplotlib in all the cases but it can be your go to tool for the plots given in examples and API in most of the cases.

Although we will use easy_mpl in this chapter but some concepts of matplotlib will be introduced along the way. Each chapter contains comprehensive list of examples so that they can be adopted by the user. For complex examples, the user will notice switching to matplotlib by using native matplotlibl functions directly.