How to use Seaborn to plot a lot of features with Python

Question

I have a Pandas DataFrame and I have a particular column that I would like to plot as a boxplot with another column made of values. The dataframe is a made of 49609 observations but the column of interest is made of 79 unique features.

Here is my attempt to do this:

sns.boxplot(x="values", y="column_of_interest",data=df)

But the axes are joined together too closely

You could try and aggregate those categories down, for example take top 20-30 by some metric of interest, and combine all others into an "other" category before plotting. — Chris Adams
– Chris Adams, Commented Mar 10, 2020 at 14:36

LTheriault · Accepted Answer · 2020-03-10 14:31:16Z

1

One of the first things you can do is call the following and play around with the height parameters to make the y-axis labels less crowded.

plt.figure(figsize=(<new_width>,<new_height>))

Depending on the characteristics of the values you're working with, it could also be helpful to develop some kind of abbreviated column that allows you to identify those unique values with less text.

Other than that, we would need to know what the axes are supposed to represent and what you're trying to learn from it to know if another visualization might be a better solution for you.

answered Mar 10, 2020 at 14:31

LTheriault

1,2408 silver badges15 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

JA-pythonista Over a year ago

I am trying to learn the price distribution. I was also thinking about a swarmplot with seaborn

LTheriault Over a year ago

In that case, I'd say that a swarm plot isn't great for this many values. In general, you're not going to find a good way to view this many visualizations in a compact space. That being said, I'd recommend using violinplots. You can read this documentation to find some variables to adjust to make each plot element more compact. From an EDA perspective rather than a python one, are there larger groups that these values can fit into?

ASH · Accepted Answer · 2020-03-13 17:51:51Z

0

It would be a lot more helpful if you posted an actual sample of your actual dataframe. Without real data, I suspect that it would probably be something like this.

data = df.groupby(['Values', 'Interest'])

Then.

sns.boxplot([data.y, data.x])

answered Mar 13, 2020 at 17:51

ASH

20.5k28 gold badges115 silver badges247 bronze badges

Collectives™ on Stack Overflow

How to use Seaborn to plot a lot of features with Python

2 Answers 2

2 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related