How to change the color of a curve when the values increase?

Question

How to change the color of a curve when the values increase?

Navigation

#1 by (1 votes)

0

I have a dataset. Sometimes some values increase compared to previous ones.

resultsCos = [(0, 0.4235497237569061), (0.005, 0.4235497237569061), (0.01, 0.4238950276243094), (0.015, 0.42382596685082874), (0.02, 0.42375690607734806), (0.025, 0.42230662983425415), (0.03, 0.4210635359116022), (0.035, 0.41671270718232045), (0.04, 0.40835635359116024), (0.045, 0.3966850828729282), (0.05, 0.3770027624309392), (0.055, 0.3546270718232044), (0.1, 0.25041436464088396), (0.06, 0.3301795580110497), (0.065, 0.30738950276243093), (0.07, 0.2892955801104972), (0.075, 0.27603591160220997), (0.08, 0.2653314917127072), (0.09, 0.2553867403314917)]

I was able to make a chart with the following code:

import matplotlib.pyplot as plt
import seaborn as sns
# plot results Cosinus similarity
target = resultsCos
zip(*target)
plt.scatter(*zip(*target))
plt.xlabel('treshold/distance')
plt.ylabel('accuracy')
plt.title('Accuracy of cosine according to the distance')
plt.tight_layout()
plt.show()

But we did not really notice when there was an improvement, and I'd like to highlight that.

How to change the color of a curve when values increase? And return them to the original color when they do not increase?

What is the method for changing the color according to the previous value?

python-3.x matplotlib

asked by anonymous 27.08.2018 / 18:58

1 answer

Django + Postgresql object has no attribute How to make edit button work in a modal window?

score 1 · Answer 1

Just use the data as a color input and change the chart's colormap . In the example below:

import numpy as np
import matplotlib.pyplot as plt

#Meus dados randomicos
x = np.random.random(10)
y = np.random.random(10)

#Mude o COLORMAP aqui
plt.cool()
plt.scatter(x, y,c=y)

Having as a result:

Inaddition,youcanhavemoreelaborateanalyzestosetthecolorofthepoints,assuggestedin comment . However, the final color change process will probably be done using% color% or by plotting the dots individually and changing the color according to its value (or leaving as is, depending on what you want to show). The two can have the same effect if applied correctly.