Skip to content

Latest commit

 

History

History
197 lines (127 loc) · 4.98 KB

DataVizPython.md

File metadata and controls

197 lines (127 loc) · 4.98 KB

Introducing Data Visualization in Python

Photo by Arif Wahid on Unsplash - https://unsplash.com/photos/y3FkHW1cyBE

Exploring data visualization in Python I came across a great course on Pluralsight. Here is a little overview and introduction on various charts and my code to example charts. Enjoy!

📄 Table of contents

"To accomplish great things we must first dream, then visualize, then plan... believe... act!" - Alfred A. Montapert

Python and it's eco system.

  • Python as a solid established programming language
  • Has many libraries (especially in the field of science and data visualization)
  • Matplotlib (data visualization library)
  • Pandas (for importing and processing data)
  • Jupyter (interactive Python Notebook)
  • Anaconda as package manager for Python

Draw an example chart:

import pandas as pd
from matplotlib import pyplot as plt

data = pd.read_csv('countries.csv')


austria = data[data.country == 'Austria']
plt.plot(austria.year, austria.gdpPerCapita)
plt.title('GDP per Capita of Austria')
plt.show()

Distribution Data with Histograms

Histograms help you understand the distribution of a numeric value in a way that you cannot with mean or median alone.

data2007 = data[data.year == 2007]
asia2007 = data2007[data2007.continent == 'Asia']
europe2007 = data2007[data2007.continent == 'Europe']

plt.subplot(211)
plt.title('Comparing GDP of EU and Asia in 2007')
plt.hist(asia2007.gdpPerCapita, 30, edgecolor='black')
plt.ylabel('Asia')
plt.subplot(212)
plt.hist(europe2007.gdpPerCapita, 30, edgecolor='black')
plt.ylabel('Europe')
plt.show()

Time Series / Line Charts

  • shows trend of over time
  • test a hypothesis on a variety of conditions
  • reduced misinterpretation of data
us = data[data.country == 'United States']
china = data[data.country == 'China']

plt.plot(us.year, us.gdpPerCapita)
plt.plot(china.year, china.gdpPerCapita)
plt.legend(['United States', 'China'])
plt.xlabel('year')
plt.ylabel('GDP per capita')
plt.show()

usGrowth = us.gdpPerCapita / us.gdpPerCapita.iloc[0] * 100
chinaGrowth = china.gdpPerCapita / china.gdpPerCapita.iloc[0] * 100

plt.plot(us.year, usGrowth)
plt.plot(china.year, chinaGrowth)
plt.legend(['United States', 'China'])
plt.xlabel('year')
plt.ylabel('GDP growth per capita')
plt.show()

Scatterplots

  • show relationships between multiple variables
  • helps you find outliners
import numpy as np

plt.scatter(data2007.gdpPerCapita, data2007.lifeExpectancy)
plt.title('GDP per capita and life expectancy in 2007')
plt.xlabel('GDP per capita')
plt.ylabel('Life expectancy')
plt.show()

plt.scatter(np.log10(data2007.gdpPerCapita), data2007.lifeExpectancy)
plt.title('GDP per capita and life expectancy in 2007')
plt.xlabel('GDP per capita - log10')
plt.ylabel('Life expectancy')
plt.show()

yearsSorted = sorted(set(data.year))

for aYear in yearsSorted:
    dataYear = data[data.year == aYear]
    plt.scatter(dataYear.gdpPerCapita, dataYear.lifeExpectancy, 5)
    plt.title(aYear)
    plt.xlim(0,60000)
    plt.ylim(25, 85)
    plt.xlabel('GDP per capita')
    plt.ylabel('Life expectancy')
    plt.show()


for aYear in yearsSorted:
    dataYear = data[data.year == aYear]
    plt.scatter(np.log10(dataYear.gdpPerCapita), dataYear.lifeExpectancy, 5)
    plt.title(aYear)
    plt.xlim(2,5)
    plt.ylim(25, 85)
    plt.xlabel('GDP per capita - log10')
    plt.ylabel('Life expectancy')
    plt.show()

Bar graphs

  • help compare numeric values
  • good for comparing multiple values
top10 = data2007.sort_values('population', ascending=False).head(10)

x = range(10)
plt.bar(x, top10.population / 10**6)
plt.xticks(x, top10.country, rotation='vertical')
plt.title('10 most populous countries')
plt.ylabel('Population in mil')
plt.show()

How to handle too much data?

  • using mean, percentiles
  • evaluating samples multiple times

Useful links & credits

pluralsight logo

See a in-depth tutorial on Pluralsight.

Thanks for reading my article! Feel free to leave any feedback!