Animated Data Scatter Diagram with Matplotlib

Question

I tried in this way to adapt the example of animated scatter, so that it shows in real time the results of the agent-based model I developed. However, the result shown in the graph is not what I except them. This happens incorrectly when updating values, and strange patterns appear where agents tend to group along a diagonal line. I have added simple code that illustrates this problem. Does anyone have an idea what will go wrong?

import matplotlib.pyplot as plt import matplotlib.animation as animation import numpy as np import os n = 25 ## nr of agents x,y = 10, 10 ## matrix of x by y dimension dataX, dataY, binaryRaster = [],[],[] class AnimatedScatter(object): """An animated scatter plot using matplotlib.animations.FuncAnimation.""" def __init__(self): global n self.numpoints = n self.stream = self.data_stream() self.fig, self.ax = plt.subplots() self.ax.set_title("My first Agent Based Model (ABM)",fontsize=14) self.ax.grid(True,linestyle='-',color='0.75') self.ani = animation.FuncAnimation(self.fig, self.update, interval=100, init_func=self.setup_plot, blit=True, repeat=False) def setup_plot(self): """Initial drawing of the scatter plot.""" global x,y dataX,dataY = next(self.stream) self.scat = self.ax.scatter(dataY, dataX, c="tomato", s=20, animated=True) self.ax.axis([0, y, x, 0]) return self.scat, def data_stream(self): """Generate a random walk (brownian motion). Data is scaled to produce a soft "flickering" effect.""" global x,y, n dataX,dataY = self.createRandomData() #printing results to ascii for validation lines = [] binaryData = np.zeros((x,y), dtype=np.int) for i in range(n): binaryData[dataX,dataY] =1 for i in range(x): line = "" for j in range(y): line += str(binaryData[i,j])+ "," line= line[:-1]+ "\n" lines.append(line) lines.append("\n") yx = np.array([dataY,dataX]) cnt = 0 while cnt < 10: dataX,dataY = self.createRandomData() yx = np.array([dataY,dataX]) #printing results to ascii for validation binaryData = np.zeros((x,y), dtype=np.int) for i in range(n): binaryData[dataX,dataY] =1 for i in range(x): line = "" for j in range(y): line += str(binaryData[i,j])+ "," line= line[:-1]+ "\n" lines.append(line) lines.append("\n") cnt+=1 yield yx #printing results to ascii for validation outNm = os.getcwd()+"\\ScatterValidation.txt" outfile = open(outNm, "w") outfile.writelines(lines) outfile.close() return def update(self, i): """Update the scatter plot.""" data = next(self.stream) self.scat.set_offsets(data[:2, :]) return self.scat, def show(self): plt.show() def createRandomData(self): """Positions n agents randomly on a raster of x by y cells. Each cell can only hold a single agent.""" global x,y,n binaryData = np.zeros((x,y), dtype=np.int) newAgents = 0 dataX,dataY = [],[] while newAgents < n: row = np.random.randint(0,x,1)[0] col = np.random.randint(0,y,1)[0] if binaryData[row][col] != 1: binaryData[row][col] = 1 newAgents+=1 for row in range(x): for col in range(y): if binaryData[row][col] == 1: dataX.append(row) dataY.append(col) return dataX, dataY def main(): global n, x, y, dataX, dataY, binaryRaster a = AnimatedScatter() a.show() return if __name__ == "__main__": main() 

Answer 1

You can fix your script in two ways, both related to changing the update function:

• Using the scatter function in the update function, clearer, I think,
• Transposing a data array before calling set_offsets in an update

Using a scatter call is the clearest fix, and you can increase the number of agents at runtime:

 def update(self, i): """Update the scatter plot.""" dataX, dataY = next(self.stream) self.scat = self.ax.scatter(dataX, dataY, c="tomato", s=20, animated=True) return self.scat, 

Transposing an array of offsets will also work:

 def update(self, i): """Update the scatter plot.""" data = next(self.stream) self.scat.set_offsets(data.transpose()) return self.scat, 

Offsets are specified as N tuples with 2 elements each, while a data array is specified as 2 tuples with N elements each, moving the data array will fix your problem.

Note. If you do not change global variables, you do not need to specify global global expressions, therefore in setup_plot , __init__ , etc. you can delete the lines global n,x,y . I would put n , x and y as instance variables of your class, and also don't need dataX , dataY and binaryRasted be defined at the top of your script.