Perceptronのパラメータ変化を見るコード

こちらです。

# -*- coding: utf-8 -*-
#
# Perceptronによる二項分類
#
# 2015/04/24 ver1.0
#

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

from numpy.random import multivariate_normal

#------------#
# Parameters #
#------------#
N1 = 30         # クラス t=1 のデータ数
Mu1 = [0,0]     # クラス t=1 の中心座標

N2 = 20         # クラス t=-1 のデータ数
Mu2 = [15,10]   # クラス t=-1 の中心座標

Variances = [20,25] # 両クラス共通の分散（2種類の分散で計算を実施）


# データセット {x_n,y_n,type_n} を用意
def prepare_dataset(variance):
    cov1 = np.array([[variance,0],[0,variance]])
    cov2 = np.array([[variance,0],[0,variance]])

    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)).reset_index()
    return df[['x','y','type']]

# Perceptronのアルゴリズム（確率的勾配降下法）を実行
def run_simulation(variance, data_graph, param_graph):
    tset = prepare_dataset(variance)
    tset1 = tset[tset['type']==1]
    tset2 = tset[tset['type']==-1]
    ymin, ymax = tset.y.min()-5, tset.y.max()+10
    xmin, xmax = tset.x.min()-5, tset.x.max()+10
    data_graph.set_ylim([ymin-1, ymax+1])
    data_graph.set_xlim([xmin-1, xmax+1])
    data_graph.scatter(tset1.x, tset1.y, marker='o')
    data_graph.scatter(tset2.x, tset2.y, marker='x')

    # パラメータの初期値とbias項の設定
    w0 = w1 = w2 = 0.1
    bias = 0.5 * (tset.x.mean() + tset.y.mean())

    # Iterationを30回実施
    paramhist = DataFrame([[w0,w1,w2]], columns=['w0','w1','w2'])
    for i in range(30):
        for index, point in tset.iterrows():
            x, y, type = point.x, point.y, point.type
            if type * (w0*bias + w1*x + w2*y) < 0:
                w0 += type * 1 
                w1 += type * x
                w2 += type * y
        paramhist = paramhist.append(
                        Series([w0,w1,w2], ['w0','w1','w2']),
                        ignore_index=True)
    # 判定誤差の計算
    err = 0
    for index, point in tset.iterrows():
        x, y, type = point.x, point.y, point.type
        if type * (w0*bias + w1*x + w2*y) < 0:
            err += 1
    err_rate = err * 100 / len(tset)

    # 結果の表示
    linex = np.arange(xmin-5, xmax+5)
    liney = - linex * w1 / w2 - bias * w0 / w2
    label = "ERR %.2f%%" % err_rate
    data_graph.plot(linex,liney,label=label,color='red')
    data_graph.legend(loc=1)
    paramhist.plot(ax=param_graph)
    param_graph.legend(loc=1)

# Main
if __name__ == '__main__':
    fig = plt.figure()
    # 2種類の分散で実行
    for c, variance in enumerate(Variances):
        subplots1 = fig.add_subplot(2,2,c+1)
        subplots2 = fig.add_subplot(2,2,c+2+1)
        run_simulation(variance, subplots1, subplots2)
    fig.show()