Sometimes, we find there is not visual output when we run the code in Jupyter notebook. To solve it, we can take advantage of the matplotlib.pyplot. Combining the application of matplotlib and seaborn, we are able to add title, axes and plt.show() to see the output. 

seaborn without a visual output?

The color in seaborn is fantastic. It supports personalized color configurations. There are many methods for color, two of them are:


* sns.color_palette(): an interface that sets colors based on the RGB principle. It accepts a palette object as a parameter, and can set the number of colors.


* sns.hls_palette(): set based on the principles of Hue, Luminance and Saturation, as its name HLS refers to. It provides evenly spaced colors.


When using these functions directly, it returns the RGB numbers for each color. For example, (0.30, 0.45, 0.70), each is within 0 and 1. To show the color, we can use sns.palplot().

University of Michigan - Ann Arbor

The experience of using seaborn can be shared here, including problems, summary, solutions and etc.

Seaborn useful functions

https://seaborn.pydata.org/introduction.html
https://python-graph-gallery.com/seaborn/

Seaborn Documentation

There are three functions in seaborn related to histograms.

sns.histplot()
sns.distplot()
sns.displot()

Notice that there are slight differences among these three!

seaborn histogram

seaborn plot functions

Pairplot can show the relationship of each feature and the distribution at the same time. 

* It may be slow to load if the dataframe is huge, since it often plots all the points.

sns.pairplot()

Color in Seaborn

sns.regplot():


Scatter plot  +  the regression line  +  the confidence interval


You can also plot with a discrete x variable showing means and confidence intervals for unique values, plot the continuous variable into discrete bins and fit a higher-order polynomial regression by setting the "order" parameter.



sns.regplot()

Heatmap is a very useful function if you want to see the correlation among the features. You can directly have an idea about the relationship from the figure in both value (annot = True) and color. 

From the example of the iris dataset, we can see that there is strong positive correlation among petal width, petal length and sepal length. 
 

sns.heatmap()

On the basis of heatmap, clustermap further mining the correlation between each line of data and clustering one by one according to the principle of minimum merge. If the correlation is tested again, the clustermap will be shown as follows. If using the orginal data, please notice that only numeric values can be used in clustermap. You should remove the other columns or change them into numbers.

sns.clustermap()

Seaborn owns some classic data sets for basic analysis and plotting. With the Internet connected, it can be obtained through the sns.load_dataset(). The returned data is Pandas.DataFrame object. There are 18 available datasets:


\['anagrams', 'anscombe', 'attention', 'brain_networks', 'car_crashes', 'diamonds', 'dots', 'exercise', 'flights', 'fmri', 'gammas', 'geyser', 'iris', 'mpg', 'penguins', 'planets', 'tips', 'titanic'\]


This name list can be obtained by using sns.get_dataset_names(). Among them, iris, tips, titanic are freguently used in various pratice and tutorials.

Learn Before

Related

Learn After