神经网络(python源代码)
神经网络的逻辑应该都是熟知的了,在这里想说明一下交叉验证
交叉验证方法:
看图大概就能理解了,大致就是先将数据集分成K份,对这K份中每一份都取不一样的比例数据进行训练和测试。得出K个误差,将这K个误差平均得到最终误差
这第一个部分是BP神经网络的建立
参数选取参照论文:基于数据挖掘技术的股价指数分析与预测研究_胡林林
- import math
- import random
- import tushare as ts
- import pandas as pd
- random.seed(0)
- def getData(id,start,end):
- df = ts.get_hist_data(id,start,end)
- DATA=pd.DataFrame(columns=['rate1', 'rate2','rate3','pos1','pos2','pos3','amt1','amt2','amt3','MA20','MA5','r'])
- P1 = pd.DataFrame(columns=['high','low','close','open','volume'])
- DATA2=pd.DataFrame(columns=['R'])
- DATA['MA20']=df['ma20']
- DATA['MA5']=df['ma5']
- P=df['close']
- P1['high']=df['high']
- P1['low']=df['low']
- P1['close']=df['close']
- P1['open']=df['open']
- P1['volume']=df['volume']
- DATA['rate1']=(P1['close'].shift(1)-P1['open'].shift(1))/P1['open'].shift(1)
- DATA['rate2']=(P1['close'].shift(2)-P1['open'].shift(2))/P1['open'].shift(2)
- DATA['rate3']=(P1['close'].shift(3)-P1['open'].shift(3))/P1['open'].shift(3)
- DATA['pos1']=(P1['close'].shift(1)-P1['low'].shift(1))/(P1['high'].shift(1)-P1['low'].shift(1))
- DATA['pos2']=(P1['close'].shift(2)-P1['low'].shift(2))/(P1['high'].shift(2)-P1['low'].shift(2))
- DATA['pos3']=(P1['close'].shift(3)-P1['low'].shift(3))/(P1['high'].shift(3)-P1['low'].shift(3))
- DATA['amt1']=P1['volume'].shift(1)/((P1['volume'].shift(1)+P1['volume'].shift(2)+P1['volume'].shift(3))/3)
- DATA['amt2']=P1['volume'].shift(2)/((P1['volume'].shift(2)+P1['volume'].shift(3)+P1['volume'].shift(4))/3)
- DATA['amt3']=P1['volume'].shift(3)/((P1['volume'].shift(3)+P1['volume'].shift(4)+P1['volume'].shift(5))/3)
- templist=(P-P.shift(1))/P.shift(1)
- tempDATA = []
- for indextemp in templist:
- tempDATA.append(1/(1+math.exp(-indextemp*100)))
- DATA['r'] = tempDATA
- DATA=DATA.dropna(axis=0)
- DATA2['R']=DATA['r']
- del DATA['r']
- DATA=DATA.T
- DATA2=DATA2.T
- DATAlist=DATA.to_dict("list")
- result = []
- for key in DATAlist:
- result.append(DATAlist[key])
- DATAlist2=DATA2.to_dict("list")
- result2 = []
- for key in DATAlist2:
- result2.append(DATAlist2[key])
- return result
- def getDataR(id,start,end):
- df = ts.get_hist_data(id,start,end)
- DATA=pd.DataFrame(columns=['rate1', 'rate2','rate3','pos1','pos2','pos3','amt1','amt2','amt3','MA20','MA5','r'])
- P1 = pd.DataFrame(columns=['high','low','close','open','volume'])
- DATA2=pd.DataFrame(columns=['R'])
- DATA['MA20']=df['ma20'].shift(1)
- DATA['MA5']=df['ma5'].shift(1)
- P=df['close']
- P1['high']=df['high']
- P1['low']=df['low']
- P1['close']=df['close']
- P1['open']=df['open']
- P1['volume']=df['volume']
- DATA['rate1']=(P1['close'].shift(1)-P1['open'].shift(1))/P1['open'].shift(1)
- DATA['rate2']=(P1['close'].shift(2)-P1['open'].shift(2))/P1['open'].shift(2)
- DATA['rate3']=(P1['close'].shift(3)-P1['open'].shift(3))/P1['open'].shift(3)
- DATA['pos1']=(P1['close'].shift(1)-P1['low'].shift(1))/(P1['high'].shift(1)-P1['low'].shift(1))
- DATA['pos2']=(P1['close'].shift(2)-P1['low'].shift(2))/(P1['high'].shift(2)-P1['low'].shift(2))
- DATA['pos3']=(P1['close'].shift(3)-P1['low'].shift(3))/(P1['high'].shift(3)-P1['low'].shift(3))
- DATA['amt1']=P1['volume'].shift(1)/((P1['volume'].shift(1)+P1['volume'].shift(2)+P1['volume'].shift(3))/3)
- DATA['amt2']=P1['volume'].shift(2)/((P1['volume'].shift(2)+P1['volume'].shift(3)+P1['volume'].shift(4))/3)
- DATA['amt3']=P1['volume'].shift(3)/((P1['volume'].shift(3)+P1['volume'].shift(4)+P1['volume'].shift(5))/3)
- templist=(P-P.shift(1))/P.shift(1)
- tempDATA = []
- for indextemp in templist:
- tempDATA.append(1/(1+math.exp(-indextemp*100)))
- DATA['r'] = tempDATA
- DATA=DATA.dropna(axis=0)
- DATA2['R']=DATA['r']
- del DATA['r']
- DATA=DATA.T
- DATA2=DATA2.T
- DATAlist=DATA.to_dict("list")
- result = []
- for key in DATAlist:
- result.append(DATAlist[key])
- DATAlist2=DATA2.to_dict("list")
- result2 = []
- for key in DATAlist2:
- result2.append(DATAlist2[key])
- return result2
- def rand(a, b):
- return (b - a) * random.random() + a
- def make_matrix(m, n, fill=0.0):
- mat = []
- for i in range(m):
- mat.append([fill] * n)
- return mat
- def sigmoid(x):
- return 1.0 / (1.0 + math.exp(-x))
- def sigmod_derivate(x):
- return x * (1 - x)
- class BPNeuralNetwork:
- def __init__(self):
- self.input_n = 0
- self.hidden_n = 0
- self.output_n = 0
- self.input_cells = []
- self.hidden_cells = []
- self.output_cells = []
- self.input_weights = []
- self.output_weights = []
- self.input_correction = []
- self.output_correction = []
- def setup(self, ni, nh, no):
- self.input_n = ni + 1
- self.hidden_n = nh
- self.output_n = no
- # init cells
- self.input_cells = [1.0] * self.input_n
- self.hidden_cells = [1.0] * self.hidden_n
- self.output_cells = [1.0] * self.output_n
- # init weights
- self.input_weights = make_matrix(self.input_n, self.hidden_n)
- self.output_weights = make_matrix(self.hidden_n, self.output_n)
- # random activate
- for i in range(self.input_n):
- for h in range(self.hidden_n):
- self.input_weights[i][h] = rand(-0.2, 0.2)
- for h in range(self.hidden_n):
- for o in range(self.output_n):
- self.output_weights[h][o] = rand(-2.0, 2.0)
- # init correction matrix
- self.input_correction = make_matrix(self.input_n, self.hidden_n)
- self.output_correction = make_matrix(self.hidden_n, self.output_n)
- def predict(self, inputs):
- # activate input layer
- for i in range(self.input_n - 1):
- self.input_cells[i] = inputs[i]
- # activate hidden layer
- for j in range(self.hidden_n):
- total = 0.0
- for i in range(self.input_n):
- total += self.input_cells[i] * self.input_weights[i][j]
- self.hidden_cells[j] = sigmoid(total)
- # activate output layer
- for k in range(self.output_n):
- total = 0.0
- for j in range(self.hidden_n):
- total += self.hidden_cells[j] * self.output_weights[j][k]
- self.output_cells[k] = sigmoid(total)
- return self.output_cells[:]
- def back_propagate(self, case, label, learn, correct):
- # feed forward
- self.predict(case)
- # get output layer error
- output_deltas = [0.0] * self.output_n
- for o in range(self.output_n):
- error = label[o] - self.output_cells[o]
- output_deltas[o] = sigmod_derivate(self.output_cells[o]) * error
- # get hidden layer error
- hidden_deltas = [0.0] * self.hidden_n
- for h in range(self.hidden_n):
- error = 0.0
- for o in range(self.output_n):
- error += output_deltas[o] * self.output_weights[h][o]
- hidden_deltas[h] = sigmod_derivate(self.hidden_cells[h]) * error
- # update output weights
- for h in range(self.hidden_n):
- for o in range(self.output_n):
- change = output_deltas[o] * self.hidden_cells[h]
- self.output_weights[h][o] += learn * change + correct * self.output_correction[h][o]
- self.output_correction[h][o] = change
- # update input weights
- for i in range(self.input_n):
- for h in range(self.hidden_n):
- change = hidden_deltas[h] * self.input_cells[i]
- self.input_weights[i][h] += learn * change + correct * self.input_correction[i][h]
- self.input_correction[i][h] = change
- # get global error
- error = 0.0
- for o in range(len(label)):
- error += 0.5 * (label[o] - self.output_cells[o]) ** 2
- return error
- def train(self, cases, labels, limit=10000, learn=0.05, correct=0.1):
- for i in range(limit):
- error = 0.0
- for i in range(len(cases)):
- label = labels[i]
- case = cases[i]
- error += self.back_propagate(case, label, learn, correct)
- def test(self,id):
- result=getData("", "2015-01-05", "2015-01-09")
- result2=getDataR("", "2015-01-05", "2015-01-09")
- self.setup(11, 5, 1)
- self.train(result, result2, 10000, 0.05, 0.1)
- for t in resulttest:
- print(self.predict(t))
下面是选取14-15年数据进行训练,16年数据作为测试集,调仓周期为20个交易日,大约1个月,对上证50中的股票进行预测,选取预测的涨幅前10的股票买入,对每只股票分配一样的资金,初步运行没有问题,但就是太慢了,等哪天有空了再运行
- import BPnet
- import tushare as ts
- import pandas as pd
- import math
- import xlrd
- import datetime as dt
- import time
- #
- #nn =BPnet.BPNeuralNetwork()
- #nn.test('000001')
- #for i in ts.get_sz50s()['code']:
- holdList=pd.DataFrame(columns=['time','id','value'])
- share=ts.get_sz50s()['code']
- time2=ts.get_k_data('')['date']
- newtime = time2[400:640]
- newcount=0
- for itime in newtime:
- print(itime)
- if newcount % 20 == 0:
- sharelist = pd.DataFrame(columns=['time','id','value'])
- for ishare in share:
- backwardtime = time.strftime('%Y-%m-%d',time.localtime(time.mktime(time.strptime(itime,'%Y-%m-%d'))-432000*4))
- trainData = BPnet.getData(ishare, '2014-05-22',itime)
- trainDataR = BPnet.getDataR(ishare, '2014-05-22',itime)
- testData = BPnet.getData(ishare, backwardtime,itime)
- try:
- print(testData)
- testData = testData[-1]
- print(testData)
- nn = BPnet.BPNeuralNetwork()
- nn.setup(11, 5, 1)
- nn.train(trainData, trainDataR, 10000, 0.05, 0.1)
- value = nn.predict(testData)
- newlist= pd.DataFrame({'time':itime,"id":ishare,"value":value},index=[""])
- sharelist = sharelist.append(newlist,ignore_index=True)
- except:
- pass
- sharelist=sharelist.sort(columns ='value',ascending=False)
- sharelist = sharelist[:10]
- holdList=holdList.append(sharelist,ignore_index=True)
- newcount+=1
- print(holdList)
神经网络(python源代码)的更多相关文章
- Python源代码目录组织结构
- Python源代码剖析笔记3-Python运行原理初探
Python源代码剖析笔记3-Python执行原理初探 本文简书地址:http://www.jianshu.com/p/03af86845c95 之前写了几篇源代码剖析笔记,然而慢慢觉得没有从一个宏观 ...
- 《python源代码剖析》笔记 Python的编译结果
本文为senlie原创.转载请保留此地址:http://blog.csdn.net/zhengsenlie 1.python的运行过程 1)对python源码进行编译.产生字节码 2)将编译结果交给p ...
- 《python源代码剖析》笔记 Python虚拟机框架
本文为senlie原创,转载请保留此地址:http://blog.csdn.net/zhengsenlie 1. Python虚拟机会从编译得到的PyCodeObject对象中依次读入每一条字节码指令 ...
- 如何打包发布加密的 Python 源代码
这里介绍一种使用 PyInstaller 和 PyArmor 来发布加密 Python 源代码的方式,能够达到以下目的 把所有 Python 源代码打包成为可执行文件,客户不需要 Python 就可以 ...
- 决策树(含python源代码)
因为最近实习的需要,所以用python里的sklearn包重新写了一次决策树 工具:sklearn,http://www.lfd.uci.edu/~gohlke/pythonlibs/#numpy:将 ...
- python 源代码分析之调试设置
首先在官方下载源代码,我下载的是最新版本3.4.3版本:https://www.python.org/ftp/python/3.4.3/Python-3.4.3.tgz 解压后的目录如下(借用网上的目 ...
- 实现一个单隐层神经网络python
看过首席科学家NG的深度学习公开课很久了,一直没有时间做课后编程题,做完想把思路总结下来,仅仅记录编程主线. 一 引用工具包 import numpy as np import matplotlib. ...
- python 源代码保护 之 xx.py -> xx.so
前情提要 之前由于项目的需要,需要我们将一部分“关键代码”隐藏起来. 虽然Python 先天支持 将源代码 编译后 生成 xxx.pyc 文件,但是破解起来相当容易 -_-!! 于是搜罗到了另外一种方 ...
随机推荐
- AndroidManifest.xml的android:name是否带.的区别
android项目里面的AndroidManifest.xml,会有这样的定义 <activity android:name=".Main" ...
- printf()输出
printf()函数是式样化输出函数, 一般用于向准则输出设备按规定式样输出消息.正在编写步骤时经常会用到此函数.printf()函数的挪用式样为: printf("<式样化字符串&g ...
- javaEE中关于dao层和services层的理解
javaEE中关于dao层和services层的理解 入职已经一个多月了,作为刚毕业的新人,除了熟悉公司的项目,学习公司的框架,了解项目的一些业务逻辑之外,也就在没学到什么:因为刚入职, 带我的那个师 ...
- 使用python+pychram进行API测试(接口测试)初级STEP 1
花了一天时间安装了解了下最基本的python+pychram进行API测试,下面这个可以指导自己以后入门:基本的开发级别还需要学习 1.python下载地址:https://www.python.or ...
- golang在linux下的开发环境部署[未完]
uname -a Linux symons_laptop 4.8.2-1-ARCH #1 SMP PREEMPT Mon Oct 17 08:11:46 CEST 2016 x86_64 GNU/Li ...
- java学习第20天(IO流)
构造方法File file = new File("e:\\demo"); 创建文件夹 File file = new File("e:\\demo"); fi ...
- NSSortDescriptor对象进行数组排序
//创建一个数组 NSArray *array = @[@"zhangsan", @"lisi", @"zhonger", @"z ...
- python容器类型:列表,字典,集合等
容器的概念我是从C++的STL中学到的 什么是容器? 容器是用来存储和组织其他对象的对象. 也就是说容器里面可以放很多东西,这些东西可以是字符串,可以是整数,可以是自定义类型,然后把这些东西有组织的存 ...
- jQuery div内容间隔1秒动态向上滚动HTML、JS代码
demo1: <!DOCTYPE html> <html> <head> <title>div内容间隔1秒动态滚动</title> < ...
- Excel实用技巧
情景:有时候,我们写了一个公式,然后想在其他行也套用这个公式,一般人都是把鼠标放在那个公式所在的单元格的右下角,然后往下拉,数据量少的时候还好,数据量大的时候就不太好操作了,此时,我们需要一个好方法. ...