BP分类预测

import pandas as pd
from sklearn.model_selection import train_test_split
from util import *
from sklearn.neural_network import MLPClassifier
#划分数据集
X_train,y_train,X_test,y_test=data_load()

# 初始化model对象
model = MLPClassifier(hidden_layer_sizes=(10, 5), max_iter=500, random_state=42)

# Fit训练
model.fit(X_train,y_train)

# 预测predict
predictions = model.predict(X_test)

#转换为一维
predictions = predictions.flatten()

#计算准确率
accuracy = (predictions == y_test).mean()
print(f"Accuracy: {accuracy}")

#绘制混淆矩阵
plot_confusion(predictions,y_test)

#绘制散点图
drawScatter([y_test, predictions], ['true', 'pred'])


# 记录每轮迭代的准确率
accuracies = []
for i in range(len(model.loss_curve_)):
    model_i = MLPClassifier(hidden_layer_sizes=(10, 5), max_iter=i+1, random_state=42)
    model_i.fit(X_train, y_train)
    predictions_i = model_i.predict(X_test)
    accuracy_i = (predictions_i == y_test).mean()
    accuracies.append(accuracy_i)

# 绘制每轮迭代的准确率图
plt.figure()
plt.plot(range(1, len(model.loss_curve_)+1), accuracies)
plt.title('每轮迭代的准确率')
plt.xlabel('迭代次数')
plt.ylabel('准确率')
plt.show()