Tensorflow2.0入门教程8：信用卡反欺诈案例

通过Keras模型类搭建神经网络模型

数据集下载：
链接：https://pan.baidu.com/s/19IQ6BHmT_zNyJoaB_pPa3g
提取码：1nfc

import tensorflow as tf
import pandas as pd
import numpy as np

pandas读取数据

data = pd.read_csv("creditcard.csv")

data.head()

	Time	V1	V2	V3	V4	V5	V6	V7	V8	V9	...	V21	V22	V23	V24	V25	V26	V27	V28	Amount	Class
0	0.0	-1.359807	-0.072781	2.536347	1.378155	-0.338321	0.462388	0.239599	0.098698	0.363787	...	-0.018307	0.277838	-0.110474	0.066928	0.128539	-0.189115	0.133558	-0.021053	149.62	0
1	0.0	1.191857	0.266151	0.166480	0.448154	0.060018	-0.082361	-0.078803	0.085102	-0.255425	...	-0.225775	-0.638672	0.101288	-0.339846	0.167170	0.125895	-0.008983	0.014724	2.69	0
2	1.0	-1.358354	-1.340163	1.773209	0.379780	-0.503198	1.800499	0.791461	0.247676	-1.514654	...	0.247998	0.771679	0.909412	-0.689281	-0.327642	-0.139097	-0.055353	-0.059752	378.66	0
3	1.0	-0.966272	-0.185226	1.792993	-0.863291	-0.010309	1.247203	0.237609	0.377436	-1.387024	...	-0.108300	0.005274	-0.190321	-1.175575	0.647376	-0.221929	0.062723	0.061458	123.50	0
4	2.0	-1.158233	0.877737	1.548718	0.403034	-0.407193	0.095921	0.592941	-0.270533	0.817739	...	-0.009431	0.798278	-0.137458	0.141267	-0.206010	0.502292	0.219422	0.215153	69.99	0

5 rows × 31 columns

数据集描述

欧洲的信用卡持卡人在2013年9月2天时间里的284807笔交易数据，其中有492笔交易是欺诈交易，占比0.172%。数据采用PCA变换映射为V1，V2，…，V28 数值型属性，只有交易时间和金额这两个变量没有经过PCA变换。输出变量为二值变量，1为欺诈，0为正常交易。

Time（交易时间，需将s转化为hh-mm-ss形式）

V1~V28（经PCA转换后的数字变量）

Amount（交易金额）

Class（交易类型，1为欺诈，0为正常交易）

df = data.drop("Time",axis=1)

df.head()

	V1	V2	V3	V4	V5	V6	V7	V8	V9	V10	...	V21	V22	V23	V24	V25	V26	V27	V28	Amount	Class
0	-1.359807	-0.072781	2.536347	1.378155	-0.338321	0.462388	0.239599	0.098698	0.363787	0.090794	...	-0.018307	0.277838	-0.110474	0.066928	0.128539	-0.189115	0.133558	-0.021053	149.62	0
1	1.191857	0.266151	0.166480	0.448154	0.060018	-0.082361	-0.078803	0.085102	-0.255425	-0.166974	...	-0.225775	-0.638672	0.101288	-0.339846	0.167170	0.125895	-0.008983	0.014724	2.69	0
2	-1.358354	-1.340163	1.773209	0.379780	-0.503198	1.800499	0.791461	0.247676	-1.514654	0.207643	...	0.247998	0.771679	0.909412	-0.689281	-0.327642	-0.139097	-0.055353	-0.059752	378.66	0
3	-0.966272	-0.185226	1.792993	-0.863291	-0.010309	1.247203	0.237609	0.377436	-1.387024	-0.054952	...	-0.108300	0.005274	-0.190321	-1.175575	0.647376	-0.221929	0.062723	0.061458	123.50	0
4	-1.158233	0.877737	1.548718	0.403034	-0.407193	0.095921	0.592941	-0.270533	0.817739	0.753074	...	-0.009431	0.798278	-0.137458	0.141267	-0.206010	0.502292	0.219422	0.215153	69.99	0

5 rows × 30 columns

任务描述：

根据V1…V28,Amount的数值，预测交易类型class(为1或者0)

pandas转numpy

all_data = np.array(df)

打乱数据

np.random.shuffle(all_data)

获取训练数据

train_data = all_data[:200000,:-1]
train_label = all_data[:200000,-1]

train_data.shape

(200000, 29)

获取测试集数据

test_data = all_data[200000:,:-1]
test_label = all_data[200000:,-1]

搭建模型

class cred_model(tf.keras.Model):
    def __init__(self):
        super().__init__()
        self.dense1 = tf.keras.layers.Dense(
            input_shape=(29,),
            units=100,
            activation=tf.nn.relu,
        )
        self.dense2 = tf.keras.layers.Dense(
            units=1,
            activation=tf.nn.sigmoid,
        )

    def call(self, inputs):
        x = self.dense1(inputs)
        output = self.dense2(x)
        return output

model = cred_model()

model.summary()

Model: "cred_model_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_2 (Dense)              multiple                  3000      
_________________________________________________________________
dense_3 (Dense)              multiple                  101       
=================================================================
Total params: 3,101
Trainable params: 3,101
Non-trainable params: 0
_________________________________________________________________

配置模型

model.compile(optimizer=tf.keras.optimizers.Adam(0.001),
             loss="binary_crossentropy",
             metrics=["acc"])

训练模型

model.fit(train_data, train_label, epochs=10, batch_size=100,
          validation_data=(test_data, test_label),)

Train on 200000 samples, validate on 84807 samples
Epoch 1/10
200000/200000 [==============================] - 4s 18us/sample - loss: 0.0240 - acc: 0.9988 - val_loss: 0.0056 - val_acc: 0.9994
Epoch 2/10
200000/200000 [==============================] - 3s 16us/sample - loss: 0.0112 - acc: 0.9992 - val_loss: 0.0075 - val_acc: 0.9994
Epoch 3/10
200000/200000 [==============================] - 4s 18us/sample - loss: 0.0091 - acc: 0.9992 - val_loss: 0.0503 - val_acc: 0.9987
Epoch 4/10
200000/200000 [==============================] - 3s 17us/sample - loss: 0.0152 - acc: 0.9993 - val_loss: 0.0097 - val_acc: 0.9993
Epoch 5/10
200000/200000 [==============================] - 3s 16us/sample - loss: 0.0108 - acc: 0.9994 - val_loss: 0.0052 - val_acc: 0.9994
Epoch 6/10
200000/200000 [==============================] - 3s 17us/sample - loss: 0.0064 - acc: 0.9994 - val_loss: 0.0076 - val_acc: 0.9992
Epoch 7/10
200000/200000 [==============================] - 3s 16us/sample - loss: 0.0101 - acc: 0.9994 - val_loss: 0.0077 - val_acc: 0.9994
Epoch 8/10
200000/200000 [==============================] - 3s 16us/sample - loss: 0.0057 - acc: 0.9994 - val_loss: 0.0070 - val_acc: 0.9994
Epoch 9/10
200000/200000 [==============================] - 3s 17us/sample - loss: 0.0075 - acc: 0.9994 - val_loss: 0.0044 - val_acc: 0.9993
Epoch 10/10
200000/200000 [==============================] - 3s 16us/sample - loss: 0.0070 - acc: 0.9994 - val_loss: 0.0053 - val_acc: 0.9994

<tensorflow.python.keras.callbacks.History at 0x16f5b4d4d30>

model.summary()

Model: "cred_model_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_2 (Dense)              multiple                  3000      
_________________________________________________________________
dense_3 (Dense)              multiple                  101       
=================================================================
Total params: 3,101
Trainable params: 3,101
Non-trainable params: 0
_________________________________________________________________