Matplotlib create real time animated graph

octain picture octain · Dec 20, 2015 · Viewed 10k times · Source

I am having a hard time setting up my code to create a real time animated graph, my code is graphing after the data is being collected, not showing every iteration. My script runs a regression function then stores in a file, then I access the files and plot them, here is what I have, what do I need to move around or change to have it graph real time? I tried moving the plot functions inside the for loop but that didn't work, any suggestions?

 fig = plt.figure()
 ax1 = fig.add_subplot(1,1,1)

 num = 10 
 for idx in range(1,num):
    c,e = Regr_magic()
        with open("CK_output.txt",'a') as CK:
            CK.write("{0},{1}\n".format(idx,c))
        with open("error_output.txt",'a') as E:
            E.write("{0},{1}\n".format(idx,e))



    def animate(i):
        pull = open('error_output.txt','r').read()
        data = pull.split('\n')
        xar = []
        yar = []

        for each in data:
            if len(each)>1:
                x,y = each.split(',')
                xar.append(float(x))
                yar.append(float(y))
            ax1.plot(xar, yar)
    ani = animation.FuncAnimation(fig, animate, interval=1000)
    plt.show()

FYI, data files contain the following, the iteration number and Ck value or error, so they look like this

1,.0554
2,.0422
3,.0553
4,.0742
5,.0232

Answer

Mike Müller picture Mike Müller · Dec 20, 2015

Solution for pre-computed results

This makes a decent animation from the data in your output file:

from matplotlib import pyplot as plt
from matplotlib import animation


fig = plt.figure()

with open('error_output.txt') as fobj:
    x, y = zip(*([float(x) for x in line.split(',')] for line in fobj))


def animate(n):
    line, = plt.plot(x[:n], y[:n], color='g')
    return line,

anim = animation.FuncAnimation(fig, animate, frames=len(x), interval=1000)
plt.show()

Solution for a real-time animation as the values are computed

Here a version that allows real-time animation of data produce by regr_magic:

import random
import time

from matplotlib import pyplot as plt
from matplotlib import animation


class RegrMagic(object):
    """Mock for function Regr_magic()
    """
    def __init__(self):
        self.x = 0
    def __call__(self):
        time.sleep(random.random())
        self.x += 1
        return self.x, random.random()

regr_magic = RegrMagic()

def frames():
    while True:
        yield regr_magic()

fig = plt.figure()

x = []
y = []
def animate(args):
    x.append(args[0])
    y.append(args[1])
    return plt.plot(x, y, color='g')


anim = animation.FuncAnimation(fig, animate, frames=frames, interval=1000)
plt.show()

The class RegrMagic is a helper the mocks Regr_magic(). The __call__method makes an instance of this class behave like a function. It has state and produces the numbers 1, 0.56565, 2, 0.65566 etc. for each call (second number is a random number). It also has a time delay to mimic the computation time.

The important thing is frames(). Replace Regr_magic() with Regr_magic() and should be good to go.

Solution for the concrete problem

A version without mocks:

import random
import time

from matplotlib import pyplot as plt
from matplotlib import animation


def frames():
    while True:
        yield Regr_magic()


fig = plt.figure()

x = []
y = []
def animate(args):
    x.append(args[0])
    y.append(args[1])
    return plt.plot(x, y, color='g')


anim = animation.FuncAnimation(fig, animate, frames=frames, interval=1000)
plt.show()