Use R’s ggplot2 Syntax in Python

Use R’s ggplot2 Syntax in Python

R language has a rich graphical tool used for producing standard professional plots called ggplot2. For someone coming from R to python, learning python’s graphical tools such as matplotlib and seaborn can be quite intimidating. For this reason, in this blog I will illustrate how, armed with R knowledge, one can proceed and create stunning plots in python without much needing the python knowledge.

Contents

• Plotnine
• Data set used
• Plotting
• Setting up your environment
• Adding Layers
• Adding Labels
• Beautify with themes
• Configure with facets

Plotnine

Plotnine is a python package that uses the grammar of graphics philosophy like that of ggplot2 in the R language. This is why implementing it will be simple and straight forward for new python users.

In order to install plotnine in your system you need to run:

pip install plotnine

Data set

The data set used in this illustration is the Texas housing data provided by the TAMU real estate center

import pandas as pd
import numpy as np
from plotnine import *
from plotnine.data import txhousing

Quick Overview of the dataset

txhousing.info()
## <class 'pandas.core.frame.DataFrame'>
## RangeIndex: 8602 entries, 0 to 8601
## Data columns (total 9 columns):
##  #   Column     Non-Null Count  Dtype  
## ---  ------     --------------  -----  
##  0   city       8602 non-null   object 
##  1   year       8602 non-null   int64  
##  2   month      8602 non-null   int64  
##  3   sales      8034 non-null   float64
##  4   volume     8034 non-null   float64
##  5   median     7986 non-null   float64
##  6   listings   7178 non-null   float64
##  7   inventory  7135 non-null   float64
##  8   date       8602 non-null   float64
## dtypes: float64(6), int64(2), object(1)
## memory usage: 605.0+ KB

From the out put above,the dataset contains 8 variables ,one variable is categorical while the rest are numeric.Other properties show that the data set doesn’t contain any missing value and has 8602 observations.

Plotting

Setting up your environment

Just like in R’s ggplot,in order to make a plot you must specify the data set you’re intending to visualize it’s variables then specify the aesthetics.The aesthetics in this case are the variables X and Y.

# few house keeping scripts
txhousing = txhousing.sort_values('month')
txhousing.month = txhousing.month.astype(object)
txhousing = txhousing[txhousing.city.isin(['Abilene','Amarillo','Arlington','Brownsville'])]
ggplot(txhousing,aes(x='month',y="sales"))
## <ggplot: (54430486)>

Adding Layers

Layers also known as geoms are added on top of the main base setup. In the code below geom_col layer is added to the base plot

A list of other usable geoms can be found here.

ggplot(txhousing,aes(x='month',y="sales"))+ \
  geom_col(fill="green")

Adding Labels

Labels help in communicating the story of data visualization process.The labels includes title, x and y axis texts.To achieve this,the labs layer is used .

ggplot(txhousing,aes(x='month',y="sales"))+ \
  geom_col(fill="green")+\
    labs(x="Month of Observation",y="Sales Observerd",title="Sales by month Analysis")

Beautify with theme

Theme is used to make necessary changes that will enhance the look of the plot. There are standard themes available but one can also use a custom theme.

To use themes, its recommenced to import the themes module form plotnine.

In the script below theme_matplotlib has been applied together with a custom theme.

from plotnine.themes import *

ggplot(txhousing,aes(x='month',y="sales"))+ \
  geom_col(fill="green")+\
    labs(x="Month of Observation",
    y="Sales Observerd",title="Sales by month Analysis")+\
    theme_matplotlib()+\
    theme(
      plot_title=element_text(size=12, face="bold",color="red"),
      axis_title_x = element_text(size=10,color="orange"),
      rect = element_rect(color="black",fill="gray"),
      plot_background=element_rect(fill='cyan'))

Configure with facets

Say we want to compare the month on month sales by countries. Faceting come in handy for achieving this.

ggplot(txhousing,aes(x='month',y="sales"))+ \
  geom_col(fill="green")+\
    labs(x="Month of Observation",
    y="Sales Observerd",title="Sales by month Analysis")+\
    theme_matplotlib()+\
    theme(
      plot_title=element_text(size=12, face="bold",color="red"),
      axis_title_x = element_text(size=10,color="orange"),
      rect = element_rect(color="black",fill="gray"),
      plot_background=element_rect(fill='cyan'))+\
    facet_wrap(' ~ city')

From the introductory examples above, its clear that with a sound experience with ggplot2, package one can quickly get up and running with visualization in python since the syntax is purely the same.

Cheers!