Perceptron Simulation Sample

perceptron_sample.py

import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from pandas import Series, DataFrame

from numpy.random import randint, randn, rand, multivariate_normal

def run_perceptron(ax):
    n1 = 60
    n2 = 40
    mu1 = [50,80]
    mu2 = [80,60]
    cov1 = np.array([[40,0],[0,20]])
    cov2 = np.array([[20,0],[0,40]])

    df1 = DataFrame(multivariate_normal(mu1,cov1,n1),columns=['x','y'])
    df1['type']=1
    df2 = DataFrame(multivariate_normal(mu2,cov2,n2),columns=['x','y'])
    df2['type']=-1
    df = pd.concat([df1,df2],ignore_index=True)
    df = df.reindex(np.random.permutation(df.index))

    ax.clear()
    ax.set_xlim(0,100)
    ax.set_ylim(0,100)
    ax.scatter(df1.x, df1.y, marker='o')
    ax.scatter(df2.x, df2.y, marker='x')

    bias = ((df.x + df.y) / 2).mean()
    dx, dy, d0 = rand()*10-20, rand()*10-20, rand()*10-20

    for i in range(100):
        err = 0
        for index, point in df.iterrows():
            x, y, type = point.x, point.y, point.type
            if type * (x * dx + y * dy + bias * d0) < 0:
                dx += type * x
                dy += type * y
                d0 += type * bias
                err += 1
        if err == 0: break

    print 'iter = %d, dx = %f, dy = %f, d0 = %f' % (i, dx, dy, d0)
    linex = np.arange(0,101)
    liney = - linex * dx / dy - (d0 * bias) / dy
    ax.plot(linex, liney)

if __name__ == '__main__':
    fig = plt.figure()
    ax = fig.add_subplot(111)
    run_perceptron(ax)
    fig.show()