CODEWITHSUNDEEP
pythonpandasmatplotlib
Jimmy K
Jimmy K

Reputation: 133

How do I add a horizontal standard deviation line in my chart?

I have a line chart showing the number of incidents every year, and a secondary line presently showing the average number of incidents. I note in my chart 2020 was particularly busy, so I'd like to add a second or third line showing the first and second standard deviation, to visualise how many standard deviations 2020 was above the average.

Below is my code but for some reason it cannot be plotted:

y = year_group['Count']
x = year_group['year']

y_mean = [np.mean(y)]*len(x)
y_std = np.std(y)

fig,ax = plt.subplots()

data_line = ax.plot(x,y, label='Data', marker='o')
mean_line = ax.plot(x,y_mean, label='Mean', linestyle='--')
std_line = ax.plot(x,y_std,label = 'Standard deviation')

plt.show()

This is the error:

ValueError: x and y must have same first dimension, but have shapes (8,) and (1,)

I understand this error but I wanted to know how to achieve my goals. This is my current output:

enter image description here

Upvotes: 1

Views: 1379

Answers (1)

Luca Clissa
Luca Clissa

Reputation: 908

You can use hlines. Here's some sample code:

import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

y = np.random.randint(40, 60, 61)
x = pd.date_range(start='2021-03-01', end='2021-04-30', freq='D')

y_mean = np.mean(y)
y_std = np.std(y)

fig,ax = plt.subplots()

data_line = ax.plot(x,y, label='Data', marker='o')
mean_line = plt.hlines(y=y_mean, xmin=x[0], xmax=x[-1], colors='green', linestyles='--', lw=2, label='Mean')
mean_line = plt.hlines(y=[y_mean-y_std, y_mean+y_std], xmin=x[0], xmax=x[-1], colors='red', linestyles='--', lw=2, label='Std Dev')

plt.show()

Edit: Expanding a bit, you're code is missing two pieces:

  • y_std is a one-element vector, here's why the error. In order to plot it you should duplicate it as you do for y_mean: y_std = [np.std(y)]*len(x)
  • also: I would guess you want to plot confidence bands, so what you should plot is y_mean +- n*y_std

Upvotes: 2

Related Questions