导入模块
from sklearn.datasets import load_iris
from sklearn.neural_network import MLPClassifier
from sklearn.model_selection import train_test_split
加载数据集
iris_data = load_iris()
拆分数据集
X_train, X_test, y_train, y_test = train_test_split(iris_data[‘data’], iris_data[‘target’], test_size=0.25, random_state=1)
print(iris_data[‘data’])
print(iris_data[‘target’])
创建神经网络模型
pf = MLPClassifier(solver=‘lbfgs’, hidden_layer_sizes=[9,7], random_state=0)
填充数据并训练
pf.fit(X_train, y_train)
评估模型
score = pf.score(X_test, y_test)
print(score)
完整代码
from sklearn.datasets import load_iris
from sklearn.neural_network import MLPClassifier
from sklearn.model_selection import train_test_split
iris_data = load_iris()
X_train, X_test, y_train, y_test = train_test_split(iris_data['data'], iris_data['target'], test_size=0.25, random_state=1)
print(iris_data['data'])
print(iris_data['target'])
pf = MLPClassifier(solver='lbfgs', hidden_layer_sizes=[9,7], random_state=0)
pf.fit(X_train, y_train)
score = pf.score(X_test, y_test)
print(score)
参数含义
-
train_data:待划分样本数据
-
train_target:待划分样本数据的结果(标签)
-
test_size:测试数据占样本数据的比例,若整数则样本数量
-
hidden_layer_sizes : 设置各隐层的结点数
-
activation : {‘identity’, ‘logistic’, ‘tanh’, ‘relu’}, 默认 ‘relu’,激活函数
identity:f(x) = x
logistic:f(x) = 1 / (1 + exp(-x))
tanh:f(x) = tanh(x)
relu:max(0, x) -
solver: {‘lbfgs’, ‘sgd’, ‘adam’}, 默认 ‘adam’,用来优化权重
lbfgs:quasi-Newton方法的优化器
sgd:随机梯度下降
adam: Kingma, Diederik, and Jimmy Ba提出的机遇随机梯度的优化器
注意:默认solver ‘adam’在相对较大的数据集上效果比较好(几千个样本或者更多), 对小数据集来说,lbfgs收敛更快效果也更好 -
alpha : float, optional, 默认0.0001,正则化参数,防止过拟合
-
random_state:随机数种子,决定随机数生成,保证每次都是同一个随机数,若为0或不填,则每次得到数据都不一样