Tensorflow入门九-keras应用计算泰坦尼克生存率（百度云源码数据资源）

上一篇：Tensorflow入门八-生成式对抗网络GAN（百度云源码数据资源）https://blog.csdn.net/qq_36187544/article/details/89919656

 

---

keras是tensorflow的高阶模块，在高版本一点的tensorflow里直接集成了keras,不用额外安装keras包。keras需要写的东西更少，开发更快，但是修改底层就更困难。高阶API

目录

资源说明

源代码

---

资源说明

链接：https://pan.baidu.com/s/1s8lks4ZOXShDA1aeSsjRig 
提取码：e44s 

data/titanic.xls是数据资源，也可以通过运行downloadData.py获取（多次执行只会输出数据已存在）

main.py是运行文件

---

源代码

利用pandas做数据处理，利用sklearn模块做标准化处理，keras构建模型，并对模型评估、保存，可视化损失函数和精确率，对Jack和Rose生存率做预测（被注释的print都可以放出来看一下效果）：

import pandas as pd
from sklearn import preprocessing
import tensorflow as tf
import matplotlib.pyplot as plt

#数据处理
#读取数据文件
from sympy import preorder_traversal

data_file_path = "data/titanic3.xls"
df_data = pd.read_excel(data_file_path)

#查看数据摘要
# print(df_data.describe())

#筛选提取特征字段
selected_cols = ['survived','name','pclass','sex','age','sibsp','parch','fare','embarked']
selected_df_data = df_data[selected_cols]

#显示缺失值的位置
# print(selected_df_data[selected_df_data.isnull().values==True])



#显示转换后前三行
# print(selected_df_data[:3])

def prepare_data(df_data):

    #删除name字段
    selected_df_data = df_data.drop(['name'],axis=1)

    # 处理缺失值
    age_mean_value = selected_df_data['age'].mean()
    selected_df_data['age'] = selected_df_data['age'].fillna(age_mean_value)
    fare_mean_value = selected_df_data['fare'].mean()
    selected_df_data['fare'] = selected_df_data['fare'].fillna(fare_mean_value)
    selected_df_data['embarked'] = selected_df_data['embarked'].fillna("S")

    # 编码转换，性别由字符串转为数字编码,港口由字母转为数字
    selected_df_data['sex'] = selected_df_data['sex'].map({'female': 0, 'male': 1}).astype(int)
    selected_df_data['embarked'] = selected_df_data['embarked'].map({'C': 0, 'Q': 1, 'S': 2}).astype(int)

    #转为ndarray数组
    ndarray_data = selected_df_data.values
    #后7列为特征值，第一列是标签值
    features = ndarray_data[:,1:]
    label = ndarray_data[:,0]

    #特征值标准化处理
    minmax_scale = preprocessing.MinMaxScaler(feature_range=(0,1))
    norm_features = minmax_scale.fit_transform(features)

    return norm_features,label

#显示标准化后前三列数据
# print(norm_features[:3])

#数据准备，打乱顺序
shuffled_df_data = selected_df_data.sample(frac=1)
x_data,y_data =prepare_data(shuffled_df_data)
train_size = int(len(x_data)*0.8)
x_train,y_train = x_data[:train_size],y_data[:train_size]
x_test,y_test = x_data[train_size:],y_data[train_size:]

#建立模型结构
model = tf.keras.models.Sequential()
#加入第一层，输入特征为7列，input_dim = 7 等价于input_shape=(7,)
model.add(tf.keras.layers.Dense(units=64,input_dim=7,use_bias=True,kernel_initializer='uniform',bias_initializer='zeros',activation='relu'))
model.add(tf.keras.layers.Dense(units=32,activation='sigmoid'))
model.add(tf.keras.layers.Dense(units=1,activation='sigmoid'))

#查看模型结果
# print(model.summary())

model.compile(optimizer=tf.keras.optimizers.Adam(0.003),loss='binary_crossentropy',metrics=['accuracy'])#metrics是训练和评估的度量值

#设置回调参数，保存模型
checkpoint_path = './checkpoint/Titanic.{epoch:02d}-{val_loss:.2f}.ckpt'
callbacks = [tf.keras.callbacks.TensorBoard(histogram_freq=2),
             tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path,save_weights_only=True,verbose=1,period=5)]


#模型训练
# train_history = model.fit(x=x_train,y=y_train,validation_split=0.2,epochs=20,batch_size=40,verbose=2)#不带回调参数
train_history = model.fit(x=x_train,y=y_train,validation_split=0.2,epochs=100,batch_size=40,callbacks=callbacks,verbose=1)

#history存储训练过程所有状态
#print(train_history.history)

#训练过程可视化
def visu_train_history(train_hostory,train_metric,validation_metric):
    plt.plot(train_history.history[train_metric])
    plt.plot(train_history.history[validation_metric])
    plt.title("Train history")
    plt.xlabel('epoch')
    plt.ylabel(train_metric)
    plt.legend(['train','validation'],loc='upper left')
    plt.show()

visu_train_history(train_history,'acc','val_acc')
visu_train_history(train_history,'loss','val_loss')

#模型评估
print("==================")
print('evaluate:')
evaluate_result = model.evaluate(x=x_test,y=y_test)
for i in range(len(evaluate_result)):
    print(model.metrics_names[i],'is',evaluate_result[i])

print('==================')
print("prediction:")
#预测
Jack_info = [0,'Jack',3,'male',23,1,0,5.000,'S']
Rose_info = [1,'Rose',1,'female',20,1,0,100.000,'S']
#创建新旅客的DataFrame
new_passenger_pd = pd.DataFrame([Jack_info,Rose_info],columns=selected_cols)
#加入信息
all_passage_pd = selected_df_data.append(new_passenger_pd)
#数据准备并计算生存率
x_features,y_label = prepare_data(all_passage_pd)
surv_probability = model.predict(x_features)
#在数据最后一列插入生存率预测
all_passage_pd.insert(len(all_passage_pd.columns),'surv_probability',surv_probability)

#显示预测结果
print(all_passage_pd[-2:])

运行结果（Rose生存率0.974297，Jack生存率 0.127492）：