Visualization Crash course

Here we mentioned the very basics for the visualization tools, just enough to understand how Pandas plotting and Seaborn are built on top of Matplotlib.

Matplotlib

It is common to create an alias for Matplotlib as plt and that will in this way:

import matplotlib.pyplot as plt

Now, since Jupyter notebooks is the most common tool it is important to mentione that we need to add an extra line after importing matplotlib.pyplot. so a common import session of a file will look like:

import numpy as np
import pandas as pd

import matplotlib,pyplot as plt

Simple plot

To simple plot we can use plot(x,y) but in jupyter notbooks we can add a ";" at the end so the matplotlib text wont be display

import numpy as np
import pandas as pd

import matplotlib,pyplot as plt

x = [0,1,2]
y = [100,200,300]

plt.plot(x,y)
#plt.plot(x,y);

in a .py file we will need to add plt.show() in order to see the graph

we will create a DataFrame that we can use to plot

housing = pd.DataFrame({'rooms':[1,1,2,2,2,3,3,3],
                       'price':[100,120,190,200,230,310,330,305]})

If we use the normal plot this will display some straight line but if we use the scatter we will have dots in the x and Y points

 plt.scatter(housing['rooms'],housing['price'])

Adding title and name to the axis

1. we draw the plot plt.plot(x,y)
2. we put the title plt.title('title')
3. we name the axis plt.xlabel('X Label'), plt.ylabel('Y Label')

plt.plot(x,y)
plt.title('Title')
plt.xlabel('X Label')
plt.ylabel('Y Label')

Adding limits or changing the axis scale

We can limit or expand the limit of the graphic, in this case we want the previous plot axis to start 0 for X and 100 for Y and finish at 2 for X and 30 for Y.

ptl.plot(x,y)

#axis and ticks
plt.xlim(0,2)
plt.ylin(100,300)

#labeling
plt.title('Title')
plt.xlabel('X Label')
plt.ylabel('Y Label')

Changing the markers

We can change the color and style of the line, but also we can change the style of the markets

ptl.plot(x,y,color='red', marker='o', markersize=20,linestyle='--')

# Axis and Ticks
plt.xlim(0,20)
plt.ylim(100,300)

#labels
plt.title('Title')
plt.xlabel('X label')
plt.ylabel('y label')

Seaborn

Seaborn is a library on top of Matplotlib that allow the creation of different charts and graphics with less code

for the examples we will use a Csv file

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

df = pd.read_csv('../DATA/heart.csv')
df.head()

Distribution plots

sns.distplot(df['age'])

Resizing and modify seaborn plots

for resizing

plt.figure(figsize=(12,8))
sns.distplot(df['age'])

to remove the KDE (Kernel Density Estimates)

sns.distplot(df['age'],kde=False)

similar to remove the histogram

sns.distplot(df['age'],hist=False)

to change the color

sns.distplot(df['age'],kde=False,bins=40,color='red')

we can limit the axis in seaborn as we limit the axis in matplotlib

sns.distplot(df['age'],kde=False,bins=40,color='green')
plt.xlim(50,60)

Count plot

From the same csv file

sns.countplot(x='sex',data=df)

sns.countplot(x='cp',data=df)

and we can use hue to add more information

sns.countplot(x='cp',data=df,hue='sex')

we can change the color ( there are predefine color colormaps)

sns.countplot(x='cp',data=df,palette='terrain')

Box Plot

sns.boxplot(x='target',y='thalach',data=df)

and in the same way that with count plots we can use the hue to add more information to the Box plot

sns.boxplot(x='target',y='thalach',data=df, hue='sex')

Scatter plot

This kind of plot is use to display the relationship between two continuous features we can use the hue and size to add extra dimension, and use palette to change the color more info

sns.scatterplot(x='chol',y='trestbps',data=df,hue='sex',size='age')

Pairplots

Pairplots perform scatterplots and histograms for every single column in your data set. This means it could be a huge plot for large datasets! Use with caution, as it could take a long time for large datasets and the figures could be too small! more info

iris = pd.read_csv('../DATA/iris.csv')
iris.head()

sns.pairplot(iris)

or just show the KDEs instead of histograms

sns.pairplot(iris, hue="species")