Python Keras搭建模型拟合数据集报错问题咨询

在将数据集拟合到构建的神经网络模型过程中遇到报错，无法理解报错含义，也不知道对应修复方法。

相关代码 #

import numpy as np
import pandas as pd
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler
from keras.models import Sequential
from keras.layers import Dense


dataset = pd.read_csv('Churn_Modelling.csv')
dataset

数据集预览 #

X=dataset.iloc[:,3:13].values
Y=dataset.iloc[:,13].values

from sklearn.preprocessing import LabelEncoder, OneHotEncoder
lableencoder_X_2 = LabelEncoder()
X[:, 2] = lableencoder_X_2.fit_transform(X[:, 2])

ct = ColumnTransformer([('ohe', OneHotEncoder(), [1])], remainder='passthrough')
X = np.array(ct.fit_transform(X), dtype = str)
X = X[:, 1:]

classifier.add(Dense(units = 6,kernel_initializer = 'uniform',activation ='relu',input_dim = 11))

classifier.add(Dense(units = 6,kernel_initializer = 'uniform',activation ='relu'))

classifier.add(Dense(units= 1, kernel_initializer = 'uniform',activation = 'sigmoid'))

classifier.compile(optimizer = 'adam', loss = 'binary_crssentropy', metrics = ['accuracy'])

# 拟合数据集到模型
classifier.fit(X_train, y_train, batch_size = 10, epochs = 100)

报错信息 #

代码中存在5处直接触发报错或导致模型无法正常训练的问题，对应修复方式如下：

• 未初始化序贯模型：代码未实例化Sequential类就直接调用classifier.add()添加网络层，会触发变量未定义报错。需要在添加网络层前先执行classifier = Sequential()完成模型初始化。
• 特征类型错误：独热编码后执行X = np.array(ct.fit_transform(X), dtype = str)强制将所有特征转为字符串格式，神经网络无法接收字符串类型的张量参与运算，需要删除dtype = str参数，保留默认数值类型即可。
• 损失函数拼写错误：编译模型时传入的损失值binary_crssentropy存在拼写错误，Keras无法识别该损失函数，正确的二分类交叉熵损失名称为binary_crossentropy。
• 缺少数据集拆分步骤：代码直接调用classifier.fit(X_train, y_train)，但从未定义X_train、y_train变量，需要先通过train_test_split将处理好的特征和标签拆分为训练集、测试集。
• 未做特征标准化：代码导入了StandardScaler但未使用，神经网络对特征尺度敏感，未缩放的特征会导致模型收敛慢、精度差，需要在数据集拆分后对特征做标准化处理。

修正后可运行完整代码 #

import numpy as np
import pandas as pd
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import LabelEncoder, OneHotEncoder, StandardScaler
from sklearn.model_selection import train_test_split
from keras.models import Sequential
from keras.layers import Dense

# 读取数据集
dataset = pd.read_csv('Churn_Modelling.csv')
X = dataset.iloc[:,3:13].values
Y = dataset.iloc[:,13].values

# 分类字段编码
labelencoder_X_2 = LabelEncoder()
X[:, 2] = labelencoder_X_2.fit_transform(X[:, 2])

ct = ColumnTransformer([('ohe', OneHotEncoder(), [1])], remainder='passthrough')
X = np.array(ct.fit_transform(X))
X = X[:, 1:] # 去除独热编码产生的冗余哑变量

# 拆分训练集、测试集
X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.2, random_state=0)

# 特征标准化
sc = StandardScaler()
X_train = sc.fit_transform(X_train)
X_test = sc.transform(X_test)

# 初始化并构建神经网络
classifier = Sequential()
classifier.add(Dense(units = 6, kernel_initializer = 'uniform', activation='relu', input_dim = 11))
classifier.add(Dense(units = 6, kernel_initializer = 'uniform', activation='relu'))
classifier.add(Dense(units= 1, kernel_initializer = 'uniform', activation='sigmoid'))

# 编译模型
classifier.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])

# 训练模型
classifier.fit(X_train, y_train, batch_size = 10, epochs = 100)

内容的提问来源于stack exchange，提问作者LidorTubul