CODEWITHSUNDEEP
pythonplotly
pr338
pr338

Reputation: 9130

How to create a boxplot not showing the outliers using Python and Plotly?

How to create a boxplot not showing the outliers using Python and Plotly?

I have a full list of points I use to create a box plot that has many outliers and the range is too big for a comparable box plot.

I just don't want to show the outliers in this list on the box plot at all.

  1. Is there a way to not show outliers in the box plot?

If not, then I tried removing the outliers from data before plotting it. However, then Plotly makes some of points that I did not remove as outliers.

  1. Is there a way to create a box plot where none of the elements are considered outliers?

Upvotes: 11

Views: 23951

Answers (2)

Tom
Tom

Reputation: 61

Late to the game but I found a (perhaps new) easy solution:

fig.update_traces(boxpoints=False)

as shown here: https://plotly.com/python/reference/box/

Note that I had to delete the selector=dict(type='box) section as it created an error.

Upvotes: 6

theengineear
theengineear

Reputation: 957

Andrew from Plotly here.

  1. You can't just not show some of the data in the array. So you can set boxpoints: "all" to get a jitter of the points, including the outliers. This will leave the boxplot as-is, without outliers sitting on top of it. I'm guessing this isn't really what you want though.

  2. To prevent outliers from being discovered in the data array, set boxpoints: false

So in Python, something like this should work:

import plotly.plotly as py
from plotly.graph_objs import Box, Figure

fig = Figure()
boxpoints_default = Box(y=[1, 2, 3, 2, 1, 10], name='default')
boxpoints_false = Box(y=[1, 2, 3, 2, 1, 10], boxpoints=False, name='no outliers')
boxpoints_all = Box(y=[1, 2, 3, 2, 1, 10], boxpoints='all', name='jitter boxpoints')

fig['data'].extend([boxpoints_default, boxpoints_false, boxpoints_all])
fig['layout'].update(title='Comparing boxplot "boxpoints" settings')

py.iplot(fig, filename='Stack Overflow 31497537')

Here's the resulting figure for that:

https://plot.ly/~theengineear/4936/comparing-boxplot-boxpoints-settings/

Here's a link to box plot tutorials in general with Plotly:

http://help.plot.ly/make-a-box-plot/

Upvotes: 11

Related Questions