Seaborn

Seaborn was

• built from matplotlib

• Integrated for pandas structures

Basic structure

sns.”chart type”(       data=”dataset”,       x=”data in x axis”,       y=”data in y axis”,       hue=”grouping variable” )

import matplotlib.pyplot as plt
import seaborn as sns

sns.barplot(x=["A", "B", "C"], y=[1,3,2])
plt.show()

---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
Cell In[1], line 1
----> 1 import matplotlib.pyplot as plt
      2 import seaborn as sns
      4 sns.barplot(x=["A", "B", "C"], y=[1,3,2])

ModuleNotFoundError: No module named 'matplotlib'

Chart types in seaborn

import seaborn as sns
import matplotlib.pyplot as plt

#loading our data
#this data represents tips in a restaurant vs other variables
tips = sns.load_dataset('tips')
print(tips.head(5))

   total_bill   tip     sex smoker  day    time  size
0       16.99  1.01  Female     No  Sun  Dinner     2
1       10.34  1.66    Male     No  Sun  Dinner     3
2       21.01  3.50    Male     No  Sun  Dinner     3
3       23.68  3.31    Male     No  Sun  Dinner     2
4       24.59  3.61  Female     No  Sun  Dinner     4

histogram (.displot())

sns.displot(data=tips, x="total_bill")
plt.show()

scatter plot (.scatterplot())

sns.scatterplot(data=tips, x="total_bill", y="tip")
plt.show()

lm plot (linear model)

this plot shows a correlation between two variables drawing a linear model on the chart

sns.lmplot(data=tips, x="total_bill", y="tip")
plt.show()

Scatter plot with group by

import seaborn as sns
import matplotlib.pyplot as plt

#import dataset
tipsdata = sns.load_dataset("tips")
tipsdata.head()

#show dataset
print(tipsdata.head())

#scatter plot, segment tip % total_bill correlation by day
sns.scatterplot(data=tipsdata, x="total_bill", y="tip", hue="day", palette ="pastel")
plt.show()

   total_bill   tip     sex smoker  day    time  size
0       16.99  1.01  Female     No  Sun  Dinner     2
1       10.34  1.66    Male     No  Sun  Dinner     3
2       21.01  3.50    Male     No  Sun  Dinner     3
3       23.68  3.31    Male     No  Sun  Dinner     2
4       24.59  3.61  Female     No  Sun  Dinner     4

heatmap

# see the correlation among the variables
tips.corr()

	total_bill	tip	size
total_bill	1.000000	0.675734	0.598315
tip	0.675734	1.000000	0.489299
size	0.598315	0.489299	1.000000

#heatmap of the correlations
sns.heatmap(tips.corr())

<AxesSubplot: >

Se pueden agregar diferentes parámetros:

• annot muestra el valor de la correlación

• cmap color

• linewidthsespacio entre variables

• linecolor color de las líneas

• vminv, max valores máximos y mínimos

• cbar=False eliminar la barra

#heatmap of the correlations
sns.heatmap(tips.corr(), annot= True, cmap='coolwarm', linewidths=5, linecolor='black',
vmin=0.5,vmax=1,cbar=False);

Kernel Density Estimation (KDE)

sns.kdeplot(data= tips, x= 'total_bill'); 

#In statistics, kernel density estimation (KDE) is the application of kernel smoothing for 
#probability density estimation, i.e., a non-parametric method to estimate the probability
#density function of a random variable based on kernels as weights.

Change chart type (kind)

Remember the image of the seaborn chart categories

you can only change to a sub category if the main category corresponds in each case.

print(tips.head(5))

   total_bill   tip     sex smoker  day    time  size
0       16.99  1.01  Female     No  Sun  Dinner     2
1       10.34  1.66    Male     No  Sun  Dinner     3
2       21.01  3.50    Male     No  Sun  Dinner     3
3       23.68  3.31    Male     No  Sun  Dinner     2
4       24.59  3.61  Female     No  Sun  Dinner     4

#Example

#lineplot is under relplot
sns.relplot(data=tips, x="total_bill", y="tip", kind="line")
plt.show()

#rugplot is under distplot
sns.displot(data=tips, x="tip", kind="ecdf")
plt.show()

Remove Legend, change palette & transparency

#note that the hue argument would add a legend of sex, but legend=False removed it.
#we also changed the line transparency with alpha=0.5
sns.displot(data= tips, x= 'total_bill', hue = 'sex', kind = 'kde', legend= False, palette='dark', alpha = .5)
plt.show()

Group by (hue)

The argument hue allows you to do a segmentation in the chart, just as group by in pandas

#let's use this data from the lesson above
print(tips.head(5))

sns.scatterplot(data=tips, x="total_bill", y="tip", hue="sex")
plt.show()

   total_bill   tip     sex smoker  day    time  size
0       16.99  1.01  Female     No  Sun  Dinner     2
1       10.34  1.66    Male     No  Sun  Dinner     3
2       21.01  3.50    Male     No  Sun  Dinner     3
3       23.68  3.31    Male     No  Sun  Dinner     2
4       24.59  3.61  Female     No  Sun  Dinner     4

Multiple charts

In this chapter see:

• how to create multiple charts one over the other

• how to create multiple charts one next to the other

import seaborn as sns
import matplotlib.pyplot as plt 

#work with the following data
tips = sns.load_dataset('tips')
tips.head(2)

	total_bill	tip	sex	smoker	day	time	size
0	16.99	1.01	Female	No	Sun	Dinner	2
1	10.34	1.66	Male	No	Sun	Dinner	3

Combine charts (overlapping)

You can combine charts literally writing one line of code below the other.

#first chart
sns.boxplot(data=tips,x="day",y="total_bill",hue="sex", dodge=True)

#second chart
sns.swarmplot(data=tips,x="day",y="total_bill",hue="sex", palette='dark:0', dodge=True)
#the dodge argument is for the swarm plot to segment by sex

plt.show()

One next to the other

the argument col separates the charts

sns.relplot(data= tips, x= 'total_bill', y = 'tip',  hue= 'day', kind= 'scatter', col = 'time');

Jointplot

Joinplot joins two different charts (not overlapping, nor one next to the other. see below for details)

import seaborn as sns
import matplotlib.pyplot as plt

#loading our data
tips = sns.load_dataset('tips')
tips.head()

#jointplot chart
sns.jointplot(data=tips, x="total_bill", y="tip", hue="sex", kind="scatter");
#with kind you edit the type of the main chart

You can add more arguments to do a better analysis

marginal_ticks

marginal_ticks creates a table for the external chart

sns.jointplot(data=tips, x="total_bill", y="tip", hue="sex", kind="kde", marginal_ticks=True);

marginal_kws

marginal_kws allos to modify determined parameters for the external chart

sns.jointplot(data=tips, x="total_bill", y="tip", hue="sex", kind="hist",
marginal_ticks=True,     #shows a small table for the external chart
marginal_kws=dict(bins= 25, fill = True, multiple= 'dodge')  #arguments only affect the external chart
)  

<seaborn.axisgrid.JointGrid at 0x7f9de75bbd30>

Modify style,pallette & font (Set)

Modify size

plt.figure(figsize=(1,1))

sns.set()
sns.barplot(x=["A", "B", "C"], y=[1,3,2])
plt.show()

Set (modify style, pallette & font

Set allows to modify:

• style and the

• font

• palette

• font scale

simultaneously

sns.set(style="dark", palette="Spectral", font_scale=3)
sns.barplot(x=["A", "B", "C"], y=[1,3,2])
plt.show()

i will restore the default styles settings by calling set() with no arguments

sns.set()
sns.barplot(x=["A", "B", "C"], y=[1,3,2])
plt.show()

Seaborn color palletes

link:

https://seaborn.pydata.org/generated/seaborn.color_palette.html#seaborn.color_palette

some examples:

sns.color_palette("husl", 9)

sns.color_palette("Spectral", as_cmap=True)

Spectral

under

bad

over

sns.color_palette("dark:#5A9_r", as_cmap=True)

blend

under

bad

over

sns.color_palette("pastel")

Seaborn themes

Link:

https://seaborn.pydata.org/generated/seaborn.set_theme.html#seaborn.set_theme

Save your chart as a png

Hola, si desean guardar los diagramas como imagen para descargarlos y usarlos en otro lado pueden usar el plt.savefig(“name.png”)

Chart customization

import seaborn as sns
import matplotlib.pyplot as plt

tipsdata = sns.load_dataset("tips")
tipsdata.head()

	total_bill	tip	sex	smoker	day	time	size
0	16.99	1.01	Female	No	Sun	Dinner	2
1	10.34	1.66	Male	No	Sun	Dinner	3
2	21.01	3.50	Male	No	Sun	Dinner	3
3	23.68	3.31	Male	No	Sun	Dinner	2
4	24.59	3.61	Female	No	Sun	Dinner	4

Acumulative charts

sns.histplot(data=tipsdata, x="tip", bins = 15, cumulative=True)
plt.show()

Statistic charts

Argument is stat() and the options are:

[‘count’, ‘density’, ‘percent’, ‘probability’ or ‘frequency’]

sns.histplot(data = tipsdata, x= 'tip', bins = 15, hue= 'sex', stat = 'density')
plt.show()

sns.histplot(data = tipsdata, x= 'tip', bins = 15, hue= 'sex', stat = 'frequency')
plt.show()

Chart grouping

Argument is multiple() and the options are:

[‘layer’, ‘stack’, ‘fill’, ‘dodge’]

#first plot with stack
sns.histplot(data = tipsdata, x= 'tip', bins = 15, hue= 'sex', multiple = 'stack')
plt.show()

#second plot with dodge
sns.histplot(data = tipsdata, x= 'tip', bins = 15, hue= 'sex', multiple = 'dodge')
plt.show()

Area below the curve

#first plot with stack
sns.kdeplot(data = tipsdata, x= 'tip', hue= 'sex', fill = True)
plt.show()

Chart types for categorical data

import seaborn as sns
import matplotlib.pyplot as plt 

tips = sns.load_dataset('tips')
tips.head(2)

#the categorical variables are "sex", "smoker", "day", "time"

	total_bill	tip	sex	smoker	day	time	size
0	16.99	1.01	Female	No	Sun	Dinner	2
1	10.34	1.66	Male	No	Sun	Dinner	3

Note that the categorical variables are “sex”, “smoker”, “day”, “time”

Catplot

catplot is useful to work with categorical data

this chart will let you to do a double group by or double segmentation

sns.catplot(data=tips, x="day", y="total_bill",hue="sex",dodge=True,kind="box",col="smoker")

plt.show()

note that the first segmentation was in “hue” and the second in “col”

“Bar plot (count)

sns.countplot(data = tips, x="day", hue="sex");

swarm plot(dots diagram)

This chart is similar to stripplot, however, this one shows better the data concentration

sns.swarmplot(data = tips, x="day", y="total_bill", hue="sex");

sns.swarmplot(data = tips, x="day", y="total_bill", hue="sex", dodge=True);
#dodge fixes the issue of one category over the other

stripplot

Looks similar to swarm plot, but data is agglomerated in this case

sns.stripplot(data = tips, x="day", y="total_bill", hue="sex", dodge=True);

boxplot separated categories

sns.boxplot(data=tips,x="day",y="total_bill",hue="sex", showfliers=True)     
#i put showfliers argument in case you want to remove outliers

plt.show()

violin plot

this plot is similar to a boxplot, but does not show quartile. It shows the data concentration

sns.violinplot(data=tips, x="day", y="total_bill")
plt.show()

Boxplot + Swarmplot

#first chart
sns.boxplot(data=tips,x="day",y="total_bill",hue="sex", dodge=True)

#second chart
sns.swarmplot(data=tips,x="day",y="total_bill",hue="sex", palette='dark:0', dodge=True, marker="<")
#the dodge argument is for the swarm plot to segment by sex

plt.show()

Correlation charts

The main chart to identify correlations is the scatter chart, this chapter will focus on this

import seaborn as sns
import matplotlib.pyplot as plt 

#data to work on
tips = sns.load_dataset('tips')
tips.head(2)

	total_bill	tip	sex	smoker	day	time	size
0	16.99	1.01	Female	No	Sun	Dinner	2
1	10.34	1.66	Male	No	Sun	Dinner	3

correlation by categories

sns.scatterplot(data= tips, x= 'total_bill', y = 'tip', hue= 'day', palette="pastel");

lm plot with multiple categories

sns.lmplot(data= tips, x= 'total_bill', y = 'tip', hue= 'day', palette="dark");

in this chart, it does not look organized because of the data. let’s see a usefull case for this chart

iris = sns.load_dataset("iris")

iris.head()

sns.lmplot(data=iris, x="sepal_length", y="petal_length", hue="species");

second segmentation in legend

with the argument style you can change the dot shape based on another category

change dot shape

this will also change the dots shape

sns.scatterplot(data= tips, x= 'total_bill', y = 'tip', hue= 'day', palette="pastel", style="time");

change dot shape (but you deciding it)

you just have to define a dictionary in which

• key is the variable name

• value is the dot shape “D” for diamond, “s” for squared, etc.

#define the shape dictionary
shapes = { "Lunch":"D", "Dinner":"s"}


sns.scatterplot(data= tips, x= 'total_bill', y = 'tip', hue= 'day', palette="pastel", style="time", 
markers=shapes);

change dot size based on numerical variable

you can change the dots size based on a numerical variable, for example *the dots are bigger if the “size” variable for this dataset is bigger

sns.scatterplot(data= tips, x= 'total_bill', y = 'tip', hue= 'day', palette="pastel", size="size")

plt.show()

Multiple correlation charts

sns.relplot(data= tips, x= 'total_bill', y = 'tip', hue= 'day', palette="pastel", size="size", col="time", kind="scatter")

plt.show()

Move the legend (relocate)

#make the chart bigger
plt.figure(figsize=(8,8))

sns.scatterplot(data= tips, x= 'total_bill', y = 'tip', hue= 'day', style="time", size="size")
plt.legend(loc="center", bbox_to_anchor=(1.2,0.5)) #bbox_to_anchor(xposition, yposition)

plt.show()

Pairplot (correlation among all the variables)

# see the correlation among the variables
tips.corr() #---> Muestra las variables correlacionadas entre si

	total_bill	tip	size
total_bill	1.000000	0.675734	0.598315
tip	0.675734	1.000000	0.489299
size	0.598315	0.489299	1.000000

This function will show you the correlation among all the numeric variables. for this particular dataset, the numeric ones are [“total_bill”, “tip”, “size”]

sns.pairplot(data=tips)
plt.show() 

Pairplot + diag_kind + hue

You can know the correlation among all the variables given a segmentation, and also change the diagonal charts.

iris = sns.load_dataset("iris")

iris.head()

sns.pairplot(data=iris, hue="species", palette="inferno", diag_kind="kde");

Pairplot corner

The argument corner eliminates the upper diagonal, avoiding repeated charts

sns.pairplot(data= tips, corner=True);

Line charts

#loading our data

import seaborn as sns
import matplotlib.pyplot as plt 

tips = sns.load_dataset('tips')
tips.head(2)

	total_bill	tip	sex	smoker	day	time	size
0	16.99	1.01	Female	No	Sun	Dinner	2
1	10.34	1.66	Male	No	Sun	Dinner	3

sns.lineplot(data=tips, x="total_bill", y="tip", hue="time", size="size", style="day");

relplot

you can do the same with relplot and modifying the chart type later on.

sns.relplot(data= tips, x= 'total_bill', y = 'tip',  hue= 'time', style= 'day', size='size', kind= 'line');

Created in Deepnote