How to plot confidence interval in matplotlib?

Question

I'm using matplotlib to plot the distribution of a data set, and want to overlay vertical lines for the confidence interval.

The density plot looks fine, but I don't see the line. Any ideas?

# Get data
import urllib.request as request
request.urlretrieve('http://seanlahman.com/files/database/baseballdatabank-master_2016-03-02.zip', "baseballdatabank-master_2016-03-02.zip")
from zipfile import ZipFile
zip = ZipFile('baseballdatabank-master_2016-03-02.zip')
zip.extractall()

import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline

batting_df = pd.read_csv("baseballdatabank-master\core\Batting.csv")
batting_df = batting_df[batting_df['AB'] > 20]
batting_df['batting_average'] = batting_df['H'] / batting_df['AB']

# Plot distribution
batting_averages = batting_df['batting_average'].dropna()
batting_averages.plot.kde()

# Plot confidence interval
import scipy.stats
import numpy as np, scipy.stats as st
stderr = st.sem(batting_averages)
interval1 = (batting_averages.mean() - stderr * 1.96, batting_averages.mean() + stderr * 1.96)
plt.plot(interval1[0], 12)
plt.show()

I'm trying to plot the vertical line at the x coordinate of the first interval, which is centered around the mean. I passed 12 as the y coordinate as this is highest value shown on the y axis.

Your y argument to plot is unitary. You'll need your kde x as the first argument, and your interval as the y... — Kartik
– Kartik, Commented Aug 18, 2016 at 2:09
Could you expand your example with the kind of data you have? — Luis
– Luis, Commented Aug 18, 2016 at 11:54
Found the answer, in that plot takes two points as arguments. — Nate Reed
– Nate Reed, Commented Aug 18, 2016 at 13:15

Luis · Accepted Answer · 2016-08-18 13:38:53Z

3

If you catch the axes of the kde plot like this:

ax = batting_averages.plot.kde()

... then you can plot vertical lines at any position you want:

stderr = st.sem(batting_averages)
ax.vlines( x=batting_averages.mean(), ymin=-1, ymax=15, color='red', label='mean' )
stderr = 0.1
ax.vlines( x=batting_averages.mean() - stderr * 1.96, ymin=-1, ymax=15, color='green', label='95% CI' )
ax.vlines( x=batting_averages.mean() + stderr * 1.96, ymin=-1, ymax=15, color='green' )
ax.set_ylim([-1,12])
ax.legend()
plt.show()

which gives you the following graph:

(note that I changed the standard error to make the lines visible)

answered Aug 18, 2016 at 13:38

Luis

3,4977 gold badges38 silver badges63 bronze badges

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

1 Comment

Nate Reed Over a year ago

Thanks, this is useful.

Community · Accepted Answer · 2017-05-23 12:25:31Z

0

Plot takes two arguments, x and y. In this case, I need to pass the x coordinates of the two points that define the line, followed by the y coordinates of the two points:

plot((x1, x2), (y1, y2))

Substituting the variables from the example above:

plt.plot((interval1[0], interval1[0]), (0, 12))
plt.plot((interval1[1], interval1[1]), (0, 12))

See: vertical & horizontal lines in matplotlib

edited May 23, 2017 at 12:25

CommunityBot

11 silver badge

answered Aug 18, 2016 at 13:20

Nate Reed

7,05115 gold badges57 silver badges71 bronze badges

Collectives™ on Stack Overflow

How to plot confidence interval in matplotlib?

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related