2

I'm interested in automatically plotting a point just above the mean peak of a distribution, represented by a kdeplot or distplot with kde. Plotting points and lines manually is simple, but I'm having difficulty deriving this maximal coordinate point.

For example, the kdeplot generated below should have a point drawn at about (3.5, 1.0):

iris = sns.load_dataset("iris")
setosa = iris.loc[iris.species == "setosa"]
sns.kdeplot(setosa.sepal_width)

enter image description here

This question is serving the ultimate goal to draw a line across to the next peak (two distributions in one graph) with a t-statistic printed above it.

Thomas Matthew
  • 2,826
  • 4
  • 34
  • 58

1 Answers1

6

Here is one way to do it. The idea here is to first extract the x and y-data of the line object in the plot. Then, get the id of the peak and finally plot the single (x,y) point corresponding to the peak of the distribution. 

import numpy as np
import seaborn as sns
iris = sns.load_dataset("iris")
setosa = iris.loc[iris.species == "setosa"]
ax = sns.kdeplot(setosa.sepal_width)

x = ax.lines[0].get_xdata() # Get the x data of the distribution
y = ax.lines[0].get_ydata() # Get the y data of the distribution
maxid = np.argmax(y) # The id of the peak (maximum of y data)
plt.plot(x[maxid],y[maxid], 'bo', ms=10)

enter image description here

Sheldore
  • 37,862
  • 7
  • 57
  • 71
  • Hi, this method only shows max of the top most point. What if I have multiple kdemaps plotted and I want max point at each? – Tarun Maganti May 17 '22 at 08:34