问题：
1. 程序添加测试集对模型进行离线验证，加上评价指标rmse
上次写的测试部分，数据来自训练集。
2. 添加网格搜索和交叉验证

RMSE函数

‘’’
from sklearn.metrics import mean_squared_error
import numpy as np
import sklearn.metrics as sm

def rmse(y_true, y_pred):
return np.sqrt(mean_squared_error(y_true, y_pred))

‘’’

需要搜索的参数：

1
2
3
4
5
6
7
8
9
10

batch_size = 5  # Low value used for live demo purposes - 100 and 1000 would be possible too, crank that up!
hidden_dim = 12  # Count of hidden neurons in the recurrent units.
layers_stacked_count = 2  # Number of stacked recurrent cells, on the neural depth axis.


learning_rate = 0.007  # Small lr helps not to diverge during training.
nb_iters = 300  # How many times we perform a training step (therefore how many times we show a batch).
lr_decay = 0.92  # default: 0.9 . Simulated annealing.
momentum = 0.5  # default: 0.0 . Momentum technique in weights update
lambda_l2_reg = 0.003  # L2 regularization of weights - avoids overfitting

交叉验证资料

k可不可以放到网格搜索里面呢？

https://zhuanlan.zhihu.com/p/24825503

需要数据

电压(mv) -56 -36 -16 4 24 44 64
时间(ms) 电导（nS） 电导（nS） 电导（nS） 电导（nS） 电导（nS） 电导（nS） 电导（nS）
0.92 0.641966251 0.90082361 1.509488212 4.654255319 16.16202946 15.537514 22.62749071
2.5 2.29273661 6.391557996 15.09488212 24.60106383 26.59574468 27.4356103 28.70651807
5 4.524333578 12.05387783 22.21104083 26.59574468 26.39116203 28.62541993 28.49544073
7.5 6.572511617 15.87165408 23.00172513 25.48758865 24.75450082 27.4356103 27.86220871
10 8.192712155 18.44543583 22.28292122 23.93617021 24.34533552 26.59574468 27.44005404
15 10.60772805 20.63315031 20.91719379 21.05496454 22.50409165 25.05599104 26.63796015
20 12.71704573 21.01921757 19.91086832 17.95212766 20.45826514 23.65621501 26.17359
30 14.67351431 21.01921757 18.68890167 15.95744681 18.20785597 21.90649496 25.32928065
40 16.20200538 21.44818119 17.53881541 15.29255319 16.16202946 20.99664054 24.8649105
50 17.02739056 21.44818119 18.25761932 15.29255319 15.54828151 20.64669653 24.48497129
60 17.11910002 21.44818119 18.25761932 15.29255319 15.34369885 20.29675252 24.10503208

问题：

该技术是由电压钳（voltageclamp）发展而来的，电压钳技术由Cole和Marment设计，后经Hodgkin和Huxley改进并成功地应用于神经纤维动作电位的研究 
 。其设计原理是根据离子作跨膜移动时形成了跨膜离子电流（I），而通透性即离子通过膜的难易程度，其膜电阻（R）的倒数，也就是膜电导（G）。因此，膜对某种离子通透性增大时，实际上时膜电阻变小，即膜对该离子的电导加大。根据欧姆定律U=IR，即I=U/R=UG，

问题：膜电导是膜电阻的倒数？

保卫细胞电信号数学模型的构建

根据保卫细胞膜电路模型

Hodgkin-Huxley 和Goldman-Hodgkin-Katz 两个方程，都是Nernst方程的推广

电流是电荷对时间的导数

我要对输入的序列进行

--- 维度扩展 
    x_test_1 = np.expand_dims(x_test[12], axis=0)


--- PCA降维
    pca = PCA(n_components=5,copy=True) 
    newX = pca.fit_transform(x)

报错提示：在post给的内容没有在excle中找到

是程序逻辑的问题，需要在看下flask

src文件夹内的文件结构如下：

├── App.vue
├── assets
│ └── logo.png
├── components
│ └── HelloWorld.vue
├── main.js
├── router.js
└── views
├── About.vue
└── Home.vue
详解：

main.js app入口点，它与根组件一起加载和初始化Vue。
app.vue 根组件，它是开始渲染所有其他组件时的起点。
‘components’ 存储UI组件
router.js 定义URL并将URL映射到对应的组件
‘views’ 存储绑定到路由器的UI组件
‘asset’ 存储静态资源，如图像和字体

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

from flask import Flask, jsonify
from flask_cors import CORS
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.naive_bayes import GaussianNB
from sklearn.decomposition import PCA
import numpy as np

# configuration
DEBUG = True

# instantiate the app
app = Flask(__name__)
app.config.from_object(__name__)

# enable CORS
CORS(app, resources={r'/*': {'origins': '*'}})


# sanity check route
@app.route('/ping', methods=['GET'])
def ping_pong():
    return jsonify('pong!,hhh')

def Bayes():
    df = pd.read_excel('./Data.xls')
    df_T=df.T
    x=df_T.iloc[:,0:1764]
    y1 = df_T.iloc[0:26,1764:1765]     #1 类的类别
    y2 = df_T.iloc[26:59,1764:1765]   #2 类的类别
    pca = PCA(n_components=5,copy=True)
    newX = pca.fit_transform(x)
    #  拆分
    newX1 = newX[0:26,:]
    newX2 = newX[26:59,:]
    x_train1,x_test1,y_train1,y_test1 = train_test_split(newX1,y1,test_size=0.2,random_state=5)
    x_train2,x_test2,y_train2,y_test2 = train_test_split(newX2,y2,test_size=0.2,random_state=5)
    x_train = np.vstack((x_train1 , x_train2))
    x_test = np.vstack((x_test1 , x_test2))
    y_train = np.vstack((y_train1 , y_train2))
    y_test = np.vstack((y_test1 , y_test2))
    print(x_test[1])
    gnb = GaussianNB()
    gnb = gnb.fit(x_train,y_train)
    return gnb

def pretreatment(list1):
    array = np.array(list1)
    return np.expand_dims(array, axis=0)

@app.route('/predict', methods=['GET'])
def main():
    gnb = Bayes()
    predict = gnb.predict(pretreatment([2187.82169246, 153.58047602, -124.83874381, -339.65293843, -5.28906774]))
    value = int(predict[0])
    return jsonify(value)

if __name__ == '__main__':
    app.run()