-
Notifications
You must be signed in to change notification settings - Fork 0
/
新数据程序4.23.txt
163 lines (141 loc) · 8.4 KB
/
新数据程序4.23.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
train_data = pd.read_csv(r"C:\Users\admin\Desktop\new_data\train_spectrum.csv", header=None)
train_data = train_data.fillna(method='pad',axis=1)
train_data=train_data.values
train_data=train_data[:,1:6144]
train_lable = pd.read_csv(r"C:\Users\admin\Desktop\new_data\train_content.csv", header=None)
train_lable = train_lable.fillna(method='pad',axis=1)
train_lable=train_lable.values
train_lable=train_lable[:,0]
row,column=np.where(np.isnan(train_data))
train_lable= np.delete(train_lable, row, axis=0)
train_data= np.delete(train_data, row, axis=0)
del_lable_0=np.where(train_lable==0)
train_lable= np.delete(train_lable, del_lable_0, axis=0)
train_data= np.delete(train_data, del_lable_0, axis=0)
row=np.where(np.isnan(train_lable))
train_lable= np.delete(train_lable, row, axis=0)
train_data= np.delete(train_data, row, axis=0)
train_data=preprocessing.minmax_scale(train_data,feature_range=(0,1),axis=1)
test_data = pd.read_csv(r"C:\Users\admin\Desktop\new_data\test_spectrum.csv", header=None)
test_data = test_data.fillna(method='pad',axis=1)
test_data=test_data.values
test_data=test_data[:,1:6144]
test_lable = pd.read_csv(r"C:\Users\admin\Desktop\new_data\test_content.csv", header=None)
test_lable = test_lable.fillna(method='pad',axis=1)
test_lable=test_lable.values[:,0]
row,column=np.where(np.isnan(test_data))
test_lable= np.delete(test_lable, row, axis=0)
test_data= np.delete(test_data, row, axis=0)
del_lable_0=np.where(test_lable==0)
test_lable= np.delete(test_lable, del_lable_0, axis=0)
test_data= np.delete(test_data, del_lable_0, axis=0)
row=np.where(np.isnan(test_lable))
test_lable= np.delete(test_lable, row, axis=0)
test_data= np.delete(test_data, row, axis=0)
test_data=preprocessing.minmax_scale(test_data,feature_range=(0,1),axis=1)
best_gamma=1.6666666666666665
best_kernel="poly"
kpca = KernelPCA(n_components=1000,fit_inverse_transform=True,kernel=best_kernel,gamma=best_gamma)
kpca.fit(train_data)
train_data=kpca.transform(train_data)
test_data=kpca.transform(test_data)
train_data=train_data.reshape(train_data.shape[0],1,1000,1)
test_data=test_data.reshape(test_data.shape[0],1,1000,1)
train_db = tf.data.Dataset.from_tensor_slices((train_data, train_lable))
train_db = train_db.shuffle(10000)
train_db = train_db.batch(100) #加入b维度
test_db = tf.data.Dataset.from_tensor_slices((test_data, test_lable))
test_db = test_db.batch(100) #加入b维度
network = Sequential([ # 封装为一个网络 layers.LeakyReLU(alpha=0.1),
layers.Conv2D(20,kernel_size=(1,3),strides=(1,1),dilation_rate=2,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,2), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,5), strides=(1,1),dilation_rate=3,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,4), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,5),strides=(1,1),dilation_rate=3,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,4), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,4), strides=(1,1),dilation_rate=1,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#ayers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,2), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,1), strides=(1,1),kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.Flatten(),
layers.Dense(1000, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
layers.BatchNormalization(),
layers.Dense(256, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
layers.BatchNormalization(),
layers.Dense(128, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
layers.BatchNormalization(),
#layers.Dense(10, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
#layers.BatchNormalization(),
layers.Dense(1, activation=None)])
early_stopping=keras.callbacks.EarlyStopping(monitor='val_loss', min_delta=0.00001,patience=60, verbose=0, mode='auto',baseline=None, restore_best_weights=True)
def step_decay(epoch):
if epoch <160:
lrate = 0.001
else:
lrate = 0.0005
return lrate
lrate = keras.callbacks.LearningRateScheduler(step_decay)
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=os.path.join("fit_logs"), histogram_freq=1)
#cmd输入tensorboard --logdir fit_logs
network.compile(optimizer = 'Adam',loss=losses.mean_squared_error)
network.fit(train_db, epochs=300,validation_data=test_db,callbacks=[early_stopping,lrate,tensorboard_callback])
mse=network.evaluate(test_db)
rmse=np.sqrt(mse)
y_predict=network.predict(test_db)
y_predict=y_predict.reshape(y_predict.shape[0])
print("真实标记：\n",test_lable)
print("预测标记：\n",y_predict)
print("真实标记与预测标记的均方根误差：",rmse)
plt.plot(test_lable,'r.',y_predict,'g.')
plt.legend(["y_true", "y_predict"])
五左右：
network = Sequential([ # 封装为一个网络 layers.LeakyReLU(alpha=0.1),
layers.Conv2D(20,kernel_size=(1,3),strides=(1,1),dilation_rate=1,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,2), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,5), strides=(1,1),dilation_rate=1,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,4), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,5),strides=(1,1),dilation_rate=1,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,4), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,4), strides=(1,1),dilation_rate=1,kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#ayers.BatchNormalization(),
layers.MaxPooling2D(pool_size=(1,2), strides=(1, 1), padding='same',data_format='channels_first'),
layers.Conv2D(8,kernel_size=(1,1), strides=(1,1),kernel_initializer='he_normal',bias_initializer='zeros',padding='same',data_format='channels_first'),
#layers.BatchNormalization(),
layers.Flatten(),
layers.Dense(1000, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
layers.BatchNormalization(),
layers.Dense(256, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
layers.BatchNormalization(),
layers.Dense(128, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
layers.BatchNormalization(),
#layers.Dense(10, activation='relu',kernel_initializer='he_normal',bias_initializer='zeros'),
#layers.BatchNormalization(),
layers.Dense(1, activation=None)])
early_stopping=keras.callbacks.EarlyStopping(monitor='val_loss', min_delta=0.00001,patience=60, verbose=0, mode='auto',baseline=None, restore_best_weights=True)
def step_decay(epoch):
if epoch <40:
lrate = 0.001
else:
lrate = 0.0005
return lrate
lrate = keras.callbacks.LearningRateScheduler(step_decay)
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=os.path.join("fit_logs"), histogram_freq=1)
#cmd输入tensorboard --logdir fit_logs
network.compile(optimizer = 'Adam',loss=losses.mean_squared_error)
network.fit(train_db, epochs=300,validation_data=test_db,callbacks=[early_stopping,lrate,tensorboard_callback])
mse=network.evaluate(test_db)
rmse=np.sqrt(mse)
y_predict=network.predict(test_db)
y_predict=y_predict.reshape(y_predict.shape[0])
print("真实标记：\n",test_lable)
print("预测标记：\n",y_predict)
print("真实标记与预测标记的均方根误差：",rmse)
plt.plot(test_lable,'r.',y_predict,'g.')
plt.legend(["y_true", "y_predict"])