-
Notifications
You must be signed in to change notification settings - Fork 0
/
code_sample.py
65 lines (51 loc) · 1.79 KB
/
code_sample.py
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
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import GRU, Dense
import numpy as np
import matplotlib.pyplot as plt
# 加载数据
x_train, y_train = ...
x_val, y_val = ...
dim = 18
# 构建模型
def build_model():
model = Sequential()
model.add(GRU(50, input_shape=(None, dim), return_sequences=True, activation="sigmoid")) # 使用50个GRU单元
model.add(Dense(1)) # 输出层,用于回归问题
return model
# 编译模型
model = build_model()
model.compile(optimizer='adam', loss="mean_squared_error")
# 训练模型
model.fit(x_train, y_train, epochs=500, batch_size=32)
# 保存权重文件、模型文件
model.save_weights('weights.h5')
model.save('model.keras')
# 预测并计算每个序列输出的和
predictions = model.predict(x_val)
# 分析准确预测和不准确预测的特征
matching_inputs = []
non_matching_inputs = []
for i in range(len(predictions)):
# 判断预测输出与真实输出是否匹配
if np.allclose(predictions[i], y_val[i], atol=0.1):
matching_inputs.append(x_val[i])
else:
non_matching_inputs.append(x_val[0][i])
matching_inputs_np = np.array(matching_inputs)
non_matching_inputs_np = np.array(non_matching_inputs)
# 可视化每个维度
for i in range(matching_inputs_np.shape[1]):
plt.figure()
plt.hist(matching_inputs_np[:, i], bins=20)
plt.title(f'输入维度 {i + 1}')
# plt.show()
plt.savefig(f"precise_prediction_chart/匹配的输入维度 {i + 1}.png")
plt.close()
for i in range(non_matching_inputs_np.shape[1]):
plt.figure()
plt.hist(non_matching_inputs_np[:, i], bins=20)
plt.title(f'输入维度 {i+1}')
# plt.show()
plt.savefig(f"precise_prediction_chart/不匹配的输入维度 {i + 1}.png")
plt.close()