shadyquant
😎
This python package allows you to quantile and plot lines where you have multiple samples, typically for visualizing uncertainty. Your data should be shaped (N, T), where N is the number of samples, T is the dimension of your lines.
Install
pip install shadyquant
Usage
Consider you have 100 lines that you want to compute confidence intervals (quantiles) on:
import matplotlib.pyplot as plt
import numpy as np
x = np.linspace(0,2,1000)
y = np.sin(x)
plt.plot(x,y)
I'll make 100 examples of this plot, each scaled by a random number.
w = np.random.normal(size=100)**2
traj = y + y * w[:, None]
plt.plot(traj.T, color='C0')
plt.show()
You can use shadyquant to plot a 95% confidence interval (default):
import sq
sq.traj_quantile(traj)
You can also do weighted quantiling, if you have weights attached to each line
sq.traj_quantile(traj, weights=w)
You may want to do fancy shading, which just plots a series of quantiles as polygons with transparency. The quantiles overlap, which gives a nice blending. The outer edges of the polygons still correspond to the 95% confidence interval.
sq.traj_quantile(traj, fancy_shading=True)
Here are some further options you can modify:
plt.figure(figsize=(8, 3))
ax = plt.gca()
sq.traj_quantile(
traj,
ax=ax,
fancy_shading=True,
lower_q_bound=1 / 3,
upper_q_bound=2 / 3,
levels=100,
color="red",
alpha=0.01,
)
4 Dec 2, 2022
162 Nov 11, 2022
2 Nov 29, 2021
4 Jun 12, 2022
0 Jul 9, 2022
38 Dec 20, 2022
97 Nov 4, 2022
30 Dec 10, 2022
80 Sep 17, 2022
1 Dec 30, 2021
394 Dec 18, 2022
138 Dec 14, 2022
1.6k Jan 6, 2023
156 Jan 3, 2023
502 Dec 28, 2022
76 Dec 15, 2022
338 Nov 25, 2022
22 Oct 31, 2022
73 Oct 02, 2022
3.4k Jan 06, 2023
90 Dec 14, 2022
2 Dec 05, 2021
2.4k Jan 04, 2023
7.1k Jan 07, 2023
14.1k Jan 03, 2023
1 Jan 05, 2022
111 Jan 03, 2023
102 Dec 24, 2022
125 Dec 24, 2022
107 Dec 26, 2022
0 Jun 25, 2022
27 Dec 03, 2022
2 Dec 26, 2021
122 Dec 21, 2022
5 Oct 10, 2022
5 Nov 18, 2022
9 Jul 15, 2022