sklearn神经网络分类

神经网络学习能力强大，在数据量足够，隐藏层足够多的情况下，理论上可以拟合出任何方程。

理论部分

sklearn提供的神经网络算法有三个:

neural_network.BernoulliRBM，neural_network.MLPClassifier，neural_network.MLPRgression

我们现在使用MLP(Multi-Layer Perception)做分类，回归其实也类似。该网络由三部分组成：输入层、隐藏层、输出层，其中隐藏层的个数可以人为设定。神经网络学习之后的知识都存在每一层的权重矩阵中，学习的过程也就是不断训练权重达到拟合的效果。权重训练比较常用的方法是反向传递(Backpropagation)

分类代码

#coding=utf-8
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn import datasets
from sklearn.neural_network import MLPClassifier
import numpy as np
from sklearn.preprocessing import StandardScalerdef main():iris = datasets.load_iris() #典型分类数据模型#这里我们数据统一用pandas处理data = pd.DataFrame(iris.data, columns=iris.feature_names)data['class'] = iris.target#这里只取两类data = data[data['class']!=2]#为了可视化方便，这里取两个属性为例scaler = StandardScaler()X = data[['sepal length (cm)','sepal width (cm)']]scaler.fit(X)#标准化数据集X = scaler.transform(X)Y = data[['class']]#划分数据集X_train, X_test, Y_train, Y_test =train_test_split(X, Y)mpl = MLPClassifier(solver='lbfgs',activation='logistic')mpl.fit(X_train, Y_train)print 'Score:\n',mpl.score(X_test, Y_test) #score是指分类的正确率#区域划分h = 0.02x_min, x_max = X[:, 0].min() - 1, X[:, 0].max() + 1y_min, y_max = X[:, 1].min() - 1, X[:, 1].max() + 1xx, yy = np.meshgrid(np.arange(x_min, x_max, h),np.arange(y_min, y_max, h))Z = mpl.predict(np.c_[xx.ravel(), yy.ravel()])Z = Z.reshape(xx.shape)plt.contourf(xx, yy, Z, cmap=plt.cm.Paired)#做出原来的散点图class1_x = X[Y['class']==0,0]class1_y = X[Y['class']==0,1]l1 = plt.scatter(class1_x,class1_y,color='b',label=iris.target_names[0])class2_x = X[Y['class']==1,0]class2_y = X[Y['class']==1,1]l2 = plt.scatter(class2_x,class2_y,color='r',label=iris.target_names[1])class3_x = X[Y['class']==2,0]class3_y = X[Y['class']==2,1]l3 = plt.scatter(class3_x,class3_y,color='g',label=iris.target_names[2])plt.legend(handles = [l1, l2,l3], loc = 'best')plt.grid(True)plt.show()if __name__ == '__main__':main()

测试结果