1 Star 0 Fork 61

生活淡笔/TDXPystock

forked from 金诺/TDXPystock 
加入 Gitee
与超过 1200万 开发者一起发现、参与优秀开源项目,私有仓库也完全免费 :)
免费加入
该仓库未声明开源许可证文件(LICENSE),使用请关注具体项目描述及其代码上游依赖。
克隆/下载
股价预测.py 11.69 KB
一键复制 编辑 原始数据 按行查看 历史
金诺 提交于 2021-06-16 16:21 . 调整项目结构
import requests
import pandas as pd
from pyecharts import options as opts
from pyecharts.charts import Kline
import webbrowser as br
import tushare as ts
import datetime
from lxml import etree
import prettytable as pt
# import matplotlib.pyplot as plt
# import statsmodels.tsa.stattools as stat
import numpy as np
# from keras.layers import LSTM, Dense
# from keras.models import Sequential
import akshare as ak #api 使用:https://akshare-4gize6tod19f2d2e-1252952517.tcloudbaseapp.com/index.html
class StockQuoteForward():
pro = ts.pro_api('d0bf482fc51bedbefa41bb38877e169a43d00bd9ebfa1f21d28151c7')
#获取日线行情数据
def getStockDatelQuote(self,stockcode,start,end):
if stockcode[0:3] == '600' or stockcode[0:2] == '68':
stockcode = stockcode + '.SH'
else:
stockcode = stockcode + '.SZ'
# start='2020-09-30'
# end=datetime.datetime.today().strftime('%Y-%m-%d')
data = self.pro.daily(ts_code=stockcode, start_date=start, end_date=end)
data = data.sort_values(by=['trade_date'], ascending=True) # 按日期升序
return data
def get_stockname(self,stockcode):
if stockcode.isdigit(): # 如果输入的是代码
stockdata = pd.DataFrame(
self.pro.stock_basic(exchange='', list_status='L', fields='ts_code,symbol,name,area,industry,list_date'))
# print(stockdata)
for stock in stockdata.iterrows():
# print(stock)
if stockcode in stock[1]['ts_code']:
print(stock[1]['name'])
# print(str(stock[1]['ts_code'])[0:6])
return str(stock[1]['name'])
# else:
continue
else:
return stockcode
###获取股票代码
def get_stockcode(self, stockname):
if stockname.isdigit(): # 如果输入的是代码
return stockname
else:
stockdata = pd.DataFrame(
self.pro.stock_basic(exchange='', list_status='L',
fields='ts_code,symbol,name,area,industry,list_date'))
# print(stockdata)
for stock in stockdata.iterrows():
# print(stock)
if stockname == stock[1]['name']:
# print(stock[1]['name'])
# print(str(stock[1]['ts_code'])[0:6])
return str(stock[1]['ts_code'])[0:6]
else:
continue
def drawLine(self,data,stockcode):
print(data)
# stockcode=self.get_stockcode()
stockname=self.get_stockname(stockcode)
kldata = data.values[:, [1,2, 3, 4, 5]] # 分别对应日期,开盘价、收盘价、最低价和最高价
print(kldata)
kdate=kldata[0]
kobj = Kline().add_xaxis(kdate)
kobj.add_yaxis(stockname+"-日K线图", kldata.tolist())
kobj.set_global_opts(
yaxis_opts=opts.AxisOpts(is_scale=True),
xaxis_opts=opts.AxisOpts(is_scale=True),
title_opts=opts.TitleOpts(title=""))
renderfile = stockcode + 'render.html'
kobj.render(renderfile)
br.open(renderfile)
# 调用的是yahoo财经,速度很慢,经常超时
def getHK_stockQuote(self,scode):
stockData=[]
table=pt.PrettyTable()
url='https://hk.finance.yahoo.com/quote/'+scode+'.HK/history?p='+scode+'.HK'
header= {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 Chrome/78.0.3904.108Safari/537.36',
}
session=requests.session()
session.headers=header
try:
response=session.get(url=url)
except BaseException as be:
print('访问错误,重试')
response = session.get(url=url)
tree = etree.HTML(response.text)
if response.status_code==200:
table_head=tree.xpath('//div[@id="app"]//*[@id="Col1-1-HistoricalDataTable-Proxy"]/section/div[2]/table/thead/tr/th//text()')
table_data=tree.xpath('//div[@id="app"]//*[@id="Col1-1-HistoricalDataTable-Proxy"]/section/div[2]/table/tbody/tr')
for tr in table_data:
# print(tr)
date=tr.xpath('./td[1]/span/text()')[0]
date =date = datetime.datetime.strptime(date,'%Y年%m月%d日').strftime('%Y-%m-%d')
open=tr.xpath('./td[2]/span/text()')[0]
high = tr.xpath('./td[3]/span/text()')[0]
low = tr.xpath('./td[4]/span/text()')[0]
close = tr.xpath('./td[5]/span/text()')[0]
adjclose=tr.xpath('./td[6]/span/text()')[0]
volumn=tr.xpath('./td[7]/span/text()')
dict={'date':date,'open':open,'high':high,'low':low,'close':close,'adjclose':adjclose,'volumn':volumn}
table.add_row([date,open,high,low,close,adjclose,volumn])
stockData.append(dict)
# print(table_head)
table.field_names = table_head # 设置表头
table.align = 'c' # 对齐方式(c:居中,l居左,r:居右)
return table_head,table,stockData #以多处形式返回,按需要接处理
else:
print('访问错误,未取到数据')
return None
#股票预测
# def QuoteForward(self,data):
# subdata = data.iloc[:-30, :4]
# for i in range(4):
# pvalue = stat.adfuller(subdata.values[:, i], 1)[1]
# print("指标 ", data.columns[i], " 单位根检验的p值为:", pvalue)
#
# subdata_diff1 = subdata.iloc[1:, :].values - subdata.iloc[:-1, :].values
# for i in range(4):
# pvalue = stat.adfuller(subdata_diff1[:, i], 1)[1]
# print("指标 ", data.columns[i], " 单位根检验的p值为:", pvalue)
#
# # 模型阶数从1开始逐一增加
# rows, cols = subdata_diff1.shape
# aicList = []
# lmList = []
#
# for p in range(1, 11):
# baseData = None
# for i in range(p, rows):
# tmp_list = list(subdata_diff1[i, :]) + list(subdata_diff1[i - p:i].flatten())
# if baseData is None:
# baseData = [tmp_list]
# else:
# baseData = np.r_[baseData, [tmp_list]]
# X = np.c_[[1] * baseData.shape[0], baseData[:, cols:]]
# Y = baseData[:, 0:cols]
# coefMatrix = np.matmul(np.matmul(np.linalg.inv(np.matmul(X.T, X)), X.T), Y)
# aic = np.log(np.linalg.det(np.cov(Y - np.matmul(X, coefMatrix), rowvar=False))) + 2 * (
# coefMatrix.shape[0] - 1) ** 2 * p / baseData.shape[0]
# aicList.append(aic)
# lmList.append(coefMatrix)
# #基于lmList[1]中获取各指标对应的线性模型,对未来30期的数据进行预测,并与验证数据集进行比较分析
# p = np.argmin(aicList) + 1
# n = rows
#
# preddf = None
# for i in range(30):
# predData = list(subdata_diff1[n + i - p:n + i].flatten())
# predVals = np.matmul([1] + predData, lmList[p - 1])
# # 使用逆差分运算,还原预测值
# predVals = data.iloc[n + i, :].values[:4] + predVals
# if preddf is None:
# preddf = [predVals]
# else:
# preddf = np.r_[preddf, [predVals]]
# # 为subdata_diff1增加一条新记录
# subdata_diff1 = np.r_[subdata_diff1, [data.iloc[n + i + 1, :].values[:4] - data.iloc[n + i, :].values[:4]]]
#
# #进一步,绘制二维图表观察预测数据与真实数据的逼近情况
# plt.figure(figsize=(10, 7))
# for i in range(4):
# plt.subplot(2, 2, i + 1)
# plt.plot(range(30), data.iloc[-30:data.shape[0], i].values, 'o-', c='black')
# plt.plot(range(30), preddf[:, i], 'o--', c='gray')
# plt.ylim(1000, 1200)
# plt.ylabel("$" + data.columns[i] + "$")
# plt.show()
# v = 100 * (1 - np.sum(np.abs(preddf - data.iloc[-30:data.shape[0], :4]).values) / np.sum(
# data.iloc[-30:data.shape[0], :4].values))
# print("Evaluation on test data: accuracy = %0.2f%% \n" % v)
#
# # 分析预测残差情况
# (np.abs(preddf - data.iloc[-30:data.shape[0], :4]) / data.iloc[-30:data.shape[0], :4]).describe()
#
# plt.figure(figsize=(10, 7))
# for i in range(4):
# plt.subplot(2, 2, i + 1)
# plt.plot(range(30), data.iloc[-30:data.shape[0], i].values, 'o-', c='black')
# plt.plot(range(30), preddf[:, i], 'o--', c='gray')
# plt.ylim(1000, 1200)
# plt.ylabel("$" + data.columns[i] + "$")
# plt.show()
# v = 100 * (1 - np.sum(np.abs(preddf - data.iloc[-30:data.shape[0], :4]).values) / np.sum(
# data.iloc[-30:data.shape[0], : 4].values))
# print("Evaluation on test data: accuracy = %0.2f%% \n" % v)
#基于LSTM算法的预测
# def LSTM_QuoteForward(self,data):
# SEQLEN = 21 #SEQLEN表示使用前期数据的长度
# dim_in = 4 #输入数据的维度
# dim_out = 4 #表示输出数据的维度
# pred_len = 30 #预测数据的长度
# vmean = data.iloc[:, :4].apply(lambda x: np.mean(x))
# vstd = data.iloc[:, :4].apply(lambda x: np.std(x))
# t0 = data.iloc[:, :4].apply(lambda x: (x - np.mean(x)) / np.std(x)).values
# X_train = np.zeros((t0.shape[0] - SEQLEN - pred_len, SEQLEN, dim_in))
# Y_train = np.zeros((t0.shape[0] - SEQLEN - pred_len, dim_out), )
# X_test = np.zeros((pred_len, SEQLEN, dim_in))
# Y_test = np.zeros((pred_len, dim_out), )
# #以上为数据初始化
# for i in range(SEQLEN, t0.shape[0] - pred_len):
# Y_train[i - SEQLEN] = t0[i]
# X_train[i - SEQLEN] = t0[(i - SEQLEN):i]
# for i in range(t0.shape[0] - pred_len, t0.shape[0]):
# Y_test[i - t0.shape[0] + pred_len] = t0[i]
# X_test[i - t0.shape[0] + pred_len] = t0[(i - SEQLEN):i]
#
# model = Sequential()
# model.add(LSTM(64, input_shape=(SEQLEN, dim_in), activation='relu', recurrent_dropout=0.01))
# model.add(Dense(dim_out, activation='linear'))
# model.compile(loss='mean_squared_error', optimizer='rmsprop')
# history = model.fit(X_train, Y_train, epochs=200, batch_size=10, validation_split=0)
# preddf = model.predict(X_test) * vstd.values + vmean.values
# #输出预测
# preddf
# preddf.shape
# #预测效果评估
# import matplotlib.pyplot as plt
# plt.figure(figsize=(10, 7))
# for i in range(4):
# plt.subplot(2, 2, i + 1)
# plt.plot(range(30), data.iloc[-30:data.shape[0], i].values, 'o-', c='black')
# plt.plot(range(30), preddf[:, i], 'o--', c='gray')
# plt.ylim(1000, 1200)
# plt.ylabel("$" + data.columns[i] + "$")
# plt.show()
# v = 100 * (1 - np.sum(np.abs(preddf - data.iloc[-30:data.shape[0], :4]).values) / np.sum(
# data.iloc[-30:data.shape[0], : 4].values))
# print("Evaluation on test data: accuracy = %0.2f%% \n" % v)
# # Evaluation on test data: accuracy = 99.01%
#获取基金排名
def get_Topjijin(self):
repsponse=ak.fund_em_open_fund_rank()
print(repsponse.head(21))
return repsponse.head(21)
if __name__ == '__main__':
sq=StockQuoteForward()
# data=sq.getStockDatelQuote('600511')
# sq.drawLine(data,'600511')
# table_head,table_data,stockData=sq.getHK_stockQuote('1765')
# print(table_data,stockData)
sq.get_Topjijin()
马建仓 AI 助手
尝试更多
代码解读
代码找茬
代码优化
Python
1
https://gitee.com/alans2014/TDXPystock.git
git@gitee.com:alans2014/TDXPystock.git
alans2014
TDXPystock
TDXPystock
master

搜索帮助